EPIGENETIC BASIS OF STEM CELL IDENTITY IN NORMAL AND MALIGNANT HEMATOPOIETIC DEVELOPMENT by Namyoung Jung

A dissertation submitted to Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy

Baltimore, Maryland July, 2015

© 2015 Namyoung Jung All Rights Reserved

Abstract

Acute myeloid leukemia (AML) is a heterogeneous hematologic malignancy characterized by subpopulations of leukemia-initiating or leukemia stem cells (LSC) that give rise to clonally related non-stem leukemic blasts. The LSC model proposes that since LSC and their blast progeny are clonally related, their functional properties must be due to epigenetic differences. In addition, the cell of origin of LSC among normal hematopoietic stem and progenitor cells (HSPCs) has yet to be clearly demonstrated. In order to investigate the role of epigenetics in LSC function and hematopoietic development, we profiled DNA methylation and gene expression of CD34+CD38-,

CD34+CD38+ and CD34- cells from 15 AML patients, along with 6 well-defined HSPC populations from 5 normal bone marrows using Illumina Infinium

HumanMethylation450 BeadChip and Affymetrix Human Genome U133 Plus 2.0 Array.

To define LSC and blast functionally, we performed engraftment assays on the three subpopulations from 15 AML patients and defined 20 LSCs and 24 blast samples. We identified the key functional LSC epigenetic signature able to distinguish LSC from blasts that consisted of 84 differential methylations regions (DMRs) in 70 genes that correlated with differential gene expression. HOXA cluster genes were enriched within the LSC epigenetic signature. We found that most of these DMRs involve epigenetic alteration independent of underlying mutations, although several are downstream targets of genetic mutation in epigenome modifying enzymes and upstream regulators. The LSC epigenetic signature could predict overall survival for AML patients independent of known risk factors such as age and cytogenetics. We characterized epigenetic changes during normal human hematopoietic development and identified key novel regulators for

ii hematopoietic differentiation such as HMHB1 and MIR539. We found that global hypomethylation is a critical mechanism of lineage commitment in human hematopoiesis.

Our DNA methylation analysis in human hematopoiesis revealed variable epigenetic regulation compared to murine hematopoiesis. Furthermore, we found that LSC populations formed two distinct clusters resembling either lymphoid-primed multipotent progenitors (L-MPPs) or granulocyte/macrophage progenitors (GMPs). These results provide the first evidence for epigenetic variation between LSC and their blast progeny in

AML, and its prognostic power. We also provided comprehensive methylome map of human hematopoiesis and identified epigenetically distinct subgroups of AML LSCs that likely reflect the cell of origin.

Readers:

Ravindra Majeti, M.D., Ph.D.

Andrew P. Feinberg, M.D., M.P.H.

Thesis Advisor:

Andrew P. Feinberg, M.D., M.P.H.

iii Acknowledgments

Over the past 6 years of my graduate study, many people have contributed to my

personal and professional growth. It would be impossible to mention everyone who

helped me to finish this long journey in graduate school, but I wish to acknowledge

couple people here.

Foremost, I would like to express my deepest gratitude to my thesis advisor, Dr.

Andrew Feinberg for his relentless support, patience, and motivation. Andy has given

insightful suggestions and comments on any subject that we had a conversion on. He also taught me how I should behave in academic setting as a professional scientist. Without

Andy’s guidance and persistent help throughout my graduate study, this thesis would not have been possible.

I would like to thank to Dr. Ravindra Majeti at Stanford who has been a wonderful thesis committee member, the second reader of my thesis, and collaborator.

Ravi has provided his expert advices and valuable knowledge in leukemia and

hematopoiesis throughout this thesis. His comments and suggestions were an enormous

help for me to learn knowledge in a different field.

I am deeply grateful to my thesis committee chair, Dr. Roger Reeves for his

generous support and encouragement for this thesis and my career, and thesis committee

member, Dr. Donald Small for his advices and support.

I wish to acknowledge Dr. Yunje Cho, my undergraduate thesis advisor, for his

encouragements and advices, which have been motivation for my PhD work.

iv I am deeply grateful to my collaborators, Dr. Andrew Gentles and Dr. Rafael

Irizarry for their statistical advices and comments, and Dr. Bo Dai for the essential part of

functional experiments of this thesis. I would like to express my gratitude toward the

AML patients and their family members for their decisions to provide valuable samples

for this thesis.

I could not have done my PhD work without the advices and supports provided by

current and former Feinberg lab members who have had impact on every aspect of this

thesis. Akiko Doi and Brian Herb have always offered great feedbacks for experiments

and encouragements to move forward during my graduate course work. It has been very

lucky to have Amy Vandiver as my bench mate for past 5 years, who has provided useful

suggestions for science, and warm support for any decision related to graduate course

work and even personal life. I’m grateful to Carolina Montano, Lindsay Rizzardi,

Hwajin Lee, Xin Li, Yun Liu, Hong Ji, Peter Murakami, Michael Multhaup, Varenka

Rodriguez, and Elisabet Pujadas for useful discussions, feedbacks, and their friendship. I thank to Rakel Tryggvadottir and Arni Runarsson for their experimental help.

I am grateful to Samsung Scholarship that has offered financial support for first five years, and Mogam Science Scholarship Foundation for financial support for last year of my graduate study.

Many friends have supported and helped me to stay centered throughout past 6 years. I have been fortunate to have warmhearted CMM classmates, who have helped me to adapt to a new culture. Friends of Korean community and in Korea have been always there to listen to my problems, and given me great emotional support.

v Last but not least, this long journey in graduate school would have been impossible without the warm support of my family. My grandparents, Jongsung Jung and

Oksoon Cho have offered priceless life lessons and taught me how to be a polite person.

My parents, Insoo Cheong and Geumsook Choi, have provided unconditional love and support for any decision that I have made. My two younger sisters, Gayoung Jung and

Chaeyoung Jung have been the best friends throughout my life giving me emotional support.

I will always be grateful to all the individuals mentioned above who shaped me as a professional scientist.

vi Table of Contents

Title Page ...... i

Abstract ...... ii

Acknowledgements ...... iv

Table of Contents ...... vii

List of Tables ...... viii

List of Figures ...... xi

Chapter 1: Introduction ...... 1

Chapter 2: Epigenetic signature of leukemia stem cell ...... 27

Chapter 3: Epigenetic basis of human normal hematopoietic development ...... 135

Chapter 4: The cell of origin of leukemia stem cell ...... 168

References ...... 190

Curriculum Vitae ...... 207

Appendix ...... See attached files

vii List of Tables

Table 1.1. French-American-British (FAB) classification ...... 25

Table 1.2. Cytogenetics and prognosis ...... 26

Table 2.1. Clinical features of AML patients in this study ...... 60

Table 2.2. Genetic mutations identified ...... 61

Table 2.3. Engraftment of AML Subpopulations ...... 62

Table 2.4. DMRs of LSC vs Blast ...... See Appendix1

Table 2.5. Summary of LSC vs Blast DMRs ...... 64

Table 2.6. LSC epigenetic signature ...... 65

Table 2.7. Second DMR analysis to examine confounding effect of MLL cases ...... 70

Table 2.8. Ingenuity pathway analysis ...... 71

Table 2.9. Ingenuity upstream regulator analysis ...... 77

Table 2.10. Association of LSC epigenetic signature with DMRs of genetic mutations ......

...... 120

Table 2.11. Multivariate analysis of overall survival of TCGA patients using either DNA methylation or gene expression ...... 125

Table 2.12. Univariate overall survival analysis for LSC epigenetic signature regarding differential gene expression in various cohorts ...... 126

viii Table 2.13. Multivariate overall survival analysis for LSC epigenetic signature regarding differential gene expression in various cohorts ...... 127

Table 2.14. Univariate overall survival analysis for genetic mutations in epigenome modifying enzymes in TCGA ...... 128

Table 2.15. Multivariate overall survival analysis including DNMT3A mutation for LSC epigenetic signature in TCGA ...... 129

Table 2.16. Multivariate overall survival analysis for LSC epigenetic signature within intermediate cytogenetic risk patients in TCGA ...... 130

Table 2.17. Antibodies for Flow Cytometry ...... 131

Table 2.18. Primers used for sequencing of TET2, IDH1, IDH2, and DNMT3A mutations of AML ...... 132

Table 3.1. Normal bone marrow donor sample analysis ...... 158

Table 3.2. Antibodies for Flow Cytometry ...... 159

Table 3.3. DMR lists for pair-wise comparisons among HSPCs ...... See Appendix2

Table 3.4. Summary of DMRs identified in the indicated pairwise comparisons ...... 161

Table 3.5. Enrichment of DMRs for normal hematopoiesis in super-enhancers ...... 162

Table 3.6. DMRs of normal hematopoiesis located in super-enhancer of different tissues and cell types ...... See Appendix3

Table 3.7. Common genes between mouse and human hematopoiesis ......

ix ...... See Appendix4

Table 3.8. Primers for bisulfite pyrosequencing ...... 165

Table 4.1. DMRs for normal hematopoiesis ...... 181

Table 4.2. FAB type distribution for L-MPP-like and GMP-like AML samples ...... 189

x List of Figures

Figure 1.1. The role of epigenetics in multicellular organism ...... 24

Figure 2.1. Pre-sort and post-sort FACS analysis of subpopulations from human

AML ...... 44

Figure 2.2. Gene expression inversely correlates with DMRs at CpG island and open sea

...... 46

Figure 2.3. Gene body methylation doesn’t show statistically significant positive correlation with gene expression ...... 48

Figure 2.4. AML LSC and Blasts exhibit epigenetic differences that define an LSC epigenetic signature ...... 51

Figure 2.5. NPM1 mutation is associated with decreased methylation and increased expression of HOXA genes ...... 52

Figure 2.6. The LSC epigenetic signature is partially dependent on underlying somatic mutations ...... 55

Figure 2.7. The LSC epigenetic signature is associated with overall survival in human

AML ...... 56

Figure 2.8. The gene expression of the LSC epigenetic signature highly correlates with clinical outcome in the TCGA dataset ...... 58

Figure 2.9. R script for multivariate survival analysis ...... 59

xi Figure 3.1. Schematic of human hematopoiesis with the immunophenotype of individual

HSPC ...... 148

Figure 3.2. Pre-sort and post-sort FACS analysis of HSPCs from human bone marrow .....

...... 149

Figure 3.3. Comprehensive DNA methylation analysis shows tight clustering of human

hematopoietic stem and progenitor cells (HSPCs) ...... 150

Figure 3.4. DMR plots indicating genomic loci for genes with previously known functions in hematopoiesis MPO and CDK6 ...... 151

Figure 3.5. DMR plots indicating genomic loci for newly identified genes with previously unknown functions in hematopoiesis HMHB1 and MIR539 ...... 152

Figure 3.6. Location of DMRs for normal hematopoiesis relative to CpG island ...... 153

Figure 3.7. Gene expression inversely correlates with DMRs at non-CpG island regions in normal hematopoiesis ...... 154

Figure 3.8. Examples of DMRs located in master transcription factor or super-enhancer ...

...... 156

Figure 3.9. Global methylation changes during hematopoietic development in human and mouse ...... 157

Figure 4.1. Epigenetic signatures define subgroups of AML LSC reflecting the cell of origin ...... 175

xii Figure 4.2. Clustering analysis of AML populations with normal HSPCs using length matched random 216 regions ...... 176

Figure 4.3. Epigenetic signatures define subgroups of AML samples in TCGA reflecting the cell of origin ...... 177

Figure 4.4. Cell identity of TCGA AML samples ...... 178

Figure 4.5. Distribution of FAB types for L-MPP-like and GMP-like TCGA AML samples ...... 179

Figure 4.6. Correlation of disease features with cell identity ...... 180

xiii Chapter1

Introduction

1 1. Epigenetics

Overview

Epigenetics is the study of information that is heritable after cell division other than the

primary DNA sequence. Epigenetics is involved in many cellular processes such as embryonic development, differentiation, and an interaction with environmental stimuli.

Mechanisms of epigenetics include DNA methylation, histone modification, chromatin factors, chromatin structure, and noncoding RNAs. Dysregulation of the epigenetic modifications is known to be involved in many diseases such as cancer.

Definition (historical and modern)

Conrad Waddington first coined the term ‘epigenetics’ in 1942 as a mechanism of

how genotype brought about phenotype during development (Waddington, 2012). He

proposed the ‘Epigenetic landscape’ as a model for cellular development that a ball is rolling down a hill to the lowest points (Goldberg et al., 2007). The ‘Epigenetic landscape’ is a metaphor of how a cell (a ball) decides its fate (the lowest points) (Goldberg et al.,

2007). The role of epigenetics is evident in multicellular organisms, where the vast

majority of cells have the same genomic sequence, yet exhibit distinct cellular

phenotypes. All cells in the human body originate from the same source, pluripotent

stem cells in the inner cell mass (ICM) of a blastocyst. The pluripotent stem cell

differentiates into different types of cells such as neurons, hepatocytes, or tubular

epithelial cells, which constitute brain, liver, and kidney, respectively. As all these cells come from the pluripotent stem cells in the ICM, they have the same genomic sequence,

2 yet display distinct phenotypes and functionality (Figure 1.1). In order to investigate the

role of epigenetics in different levels of biological processes systematically, NIH

Roadmap Epigenomics Project was launched in 2008 (Bernstein et al., 2010). The NIH

Roadmap Epigenomics Project defines epigenetics as both heritable and other stable and long-term changes in gene expression regulation of a cell. Collective efforts to provide a public resource of epigenomic information have generated comprehensive epigenomic maps of DNA methylation, histone modifications, DNA binding protein, chromatin accessibility, and noncoding RNAs in diverse cell types and tissues (Bernstein et al.,

2010; Roadmap Epigenomics et al., 2015). Among the epigenetic modifications, DNA methylation will be the focus of this thesis.

DNA methylation

In mammals, DNA methylation is generally referred to as a covalent modification in which a methyl group is attached to the fifth position of cytosine in CpG dinucleotides.

Note that non-CG methylation in CHG and CHH (H=A, C or T) contexts has been reported in stem cell, neuron, and other differentiated tissues, yet its functional role is still an active research area (Lister et al., 2013; Lister et al., 2009; Ramsahoye et al., 2000;

Schultz et al., 2015). DNA methylation is established and maintained by three DNA methyltransferases: DNMT1, DNMT3A and DNMT3B (Li et al., 1992; Okano et al.,

1999). These enzymes use S-adenosyl methionine (SAM) as a methyl group donor.

DNMT1 is responsible for methylation of hemimethylated DNA after DNA replication, while DNMT3A and DNMT3B are able to methylate both hemi and unmethylated DNA,

3 so serve as de novo methyl transferases (Leonhardt et al., 1992; Okano et al., 1999). The

human genome contains about 28 million CpGs and 60-80% of those are generally

methylated (Smith and Meissner, 2013). A small portion (~10%) of CpGs, not

methylated clusters together, and thus establishes a genomic region called CpG islands,

which are predominantly located in the promoter of coding genes (Bird, 1986; Deaton

and Bird, 2011). Maintaining the unmethylated status of CpG islands requires

transcription factor binding, deposition of histone variant H2A.Z, and trimethylation of

histone H3 at lysine4 (H3K4me3) that inhibit the binding of DNA methyltransferases

(Brandeis et al., 1994; Conerly et al., 2010; Macleod et al., 1994; Otani et al., 2009). The

repression of CpG island-promoters genes is largely mediated by histone modification,

particularly trimethylation of histone H3 at lysine27 (H3K27me3), and polycomb

proteins (Bartke et al., 2010; Brinkman et al., 2012; Jones, 2012). However, DNA

methylation at CpG island promoters does occur to suppress CpG island containing genes that are targets of specific biological processes such as X-chromosome inactivation and genomic imprinting (Jones, 2012).

Recently, the CpG-island centric view has been challenged, as methods to measure DNA methylation have been improved with the development of array and sequencing technology. There has been an accumulation of evidence that other genomic regions also have important roles in development and disease progression. A large portion (~80%) of methylation differences among different tissues in human and mouse has been reported to not occur in CpG islands, but ‘CpG island shores’, located up to 2kb distant from CpG islands (Irizarry et al., 2009). Besides the differential methylation involved in normal tissue development, most differentially methylated regions (DMRs)

4 that distinguish human colon cancer from normal colon tissues, human induced

pluripotent stem (iPS) cells from fibroblast, and murine hematopoietic stem and

progenitor cells (HSPCs) are located in CpG island shores (Doi et al., 2009; Hansen et al.,

2011; Irizarry et al., 2009; Ji et al., 2010). Interestingly, the tissue-specific, cancer-

specific and reprogramming-specific DMRs (tDMR, cDMR, and rDMR, respectively)

showed statistically significant overlap with each other, indicating that there may be a

core set of genomic regions targeted for epigenetic regulation during normal development

and disease progression. In addition, DNA methylation at CpG island shores showed

strong inverse correlation with gene expression, while DNA methylation at CpG islands

did not show a statistically significant correlation with gene expression in these data sets,

suggesting functional importance of CpG island shores in diverse biological contexts

(Doi et al., 2009; Irizarry et al., 2009; Ji et al., 2010).

In cancer epigenetics, it has long been known that hypermethylation in CpG

islands is a core mechanism of cancer progression, yet various types of alterations have

been observed in other genomic regions. For example, in colon cancer, hypermethylation

is a major type of change in CpG island shores (Irizarry et al., 2009). A large-scale

change in half the genomic regions called hypomethylated ‘block’ has been identified in solid tumors with increased variation in expression of genes inside the blocks (Hansen et al., 2011; Timp et al., 2014; Timp and Feinberg, 2013). Stochastic variation of DNA methylation in cDMRs distinguished solid tumors from its normal counter parts, and different stages of tumors (Hansen et al., 2011; Timp et al., 2014). Thus, epigenetic dysregulation of increased epigenetic plasticity with genetic mutation would be a critical mechanism of cancer progression (Feinberg et al., 2006; Timp and Feinberg, 2013).

5 Recently, this hypomethylated large domain has also been observed in ageing phenotype.

Sun-exposed epidermal samples in older people (Age>65) showed the hypomethylated

blocks that overlapped with blocks in colon cancer and squamous cell carcinoma

(Vandiver et al., 2015). The hypomethylated blocks largely overlap with higher order

genomic regions such as large organized chromatin lysine-modifications (LOCKs) or

nuclear lamin-associated domains (LADs), suggesting the altered large scale DNA

methylation is associated with dysregulation of the large scale genomic regions in disease

progression.

Besides CpG islands and shores, gene body methylation has received a lot of

attention, yet its functional importance remains to be determined. DNA methylation in

gene bodies has traditionally been thought to silence repetitive DNA sequences, such as

retroviruses and LINE elements (Yoder et al., 1997). Recent evidence suggests that gene

body methylation positively correlates with gene expression in normal tissues and cancer samples (Kulis et al., 2012; Varley et al., 2013; Yang et al., 2014). Potential

mechanisms of gene body methylation in the regulation of gene expression include

effects on transcription elongation or splicing regulation (Jones, 2012; Laurent et al.,

2010). However, the functional role of gene body methylation in transcription regulation

is not evident yet, compared to other genomic regions.

DNA demethylation

DNA methylation is a stable covalent modification on genomic sequences, while

it is also reversible. Both passive and active loss of DNA methylation can occur through

6 diverse biological processes. Passive DNA demethylation occurs during consecutive

DNA replication in the absence of functional DNMT1 activity (Kohli and Zhang, 2013).

Recent advances in identification of enzymes involved in methyl group removal from

cytosine have facilitated our understanding of active DNA demethylation process. Ten-

eleven translocation (TET) family enzymes, TET1, TET2, and TET3 are able to oxidize

methyl cytosine and generate intermediate products including 5-hydroxymethylcytosine

(5hmC), 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC) (Ito et al., 2010; Ito et al.,

2011; Kriaucionis and Heintz, 2009; Tahiliani et al., 2009). These oxidized 5mC intermediates undergo further removal processes: passive removal by sequential DNA replication, direct removal, or DNA repair pathway-associated removal (Kohli and Zhang,

2013). Among these, the base excision repair (BER) pathway has been actively investigated. Thymine DNA glycosylase (TDG), an enzyme involved in BER is known to have an ability to remove thymine from G-T mismatches from normal DNA context

(Cortazar et al., 2007). Recently, it has been reported that TDG is required for epigenetic stability in embryonic development in mice (Cortellino et al., 2011). Since TDG has an ability to remove thymine from G-T mispair, it has been hypothesized that 5mC or oxidized 5mCs may be converted to thymine or uracil by deaminase first. Several studies have suggested that AID/APOBEC enzymes, known cytosine deaminases, play a role in deamination of the 5mC in reprogramming or embryonic development (Bhutani et al.,

2010; Kumar et al., 2013; Popp et al., 2010). Yet, a controversy over the role of the deaminases in DNA demethylation still exists, due to their limited enzyme activities on modified cytosines (Kohli and Zhang, 2013). In addition to the deaminase-mediated BER, other studies have shown that TDG can directly remove 5fC and 5caC (He et al., 2011;

7 Maiti and Drohat, 2011). DNA demethylation is implicated in multiple biological

processes including pre-implantation methylation dynamics, primordial germ cell (PGS)

reprograming, maintenance of stem cell pluripotency and cancer development (Kohli and

Zhang, 2013).

Methods to measure DNA methylation

Array-based (CHARM and Illumina Infinium HumanMethylation450 BeadChip)

In 2008, the comprehensive high-throughput arrays for relative methylation

(CHARM) method was developed to provide the first platform to interrogate DNA methylation in a genome-wide and non-CpG island biased manner (Irizarry et al., 2008).

CHARM utilizes a methylation dependent restriction enzyme, McrBC, that cleaves DNA containing methylated cytosines. Sheared genomic DNA is divided into two fractions; one for undigested control and the other for McrBC digestion. The undigested control and

McrBC digested samples undergo size selection. The size selected DNA is amplified and hybridized to a tiling array that excluded isolated CpGs. The log-ratio of the signal intensities from the array of the untreated and McrBC treated samples (M-value) is measured. Since methylation status of neighboring CpGs is likely to correlate with each other, the measured M-value is averaged within a given genomic region of interest. This process is called ‘Genome-weighted smoothing’ that improves the accuracy and specificity of measuring methylation at CpG sites (Irizarry et al., 2008). Many studies have applied CHARM to identify genome-wide differential methylation in different model systems: tissue development, cellular reprogramming, hematopoiesis, cancer, and

8 behavior of social insects (Doi et al., 2009; Herb et al., 2012; Irizarry et al., 2009; Ji et al.,

2010; Kim et al., 2010; Kim et al., 2011).

Illumina has endeavored to develop commercially available arrays to measure genome-wide DNA methylation, and produced Infinium HumanMethylation27 BeadChip

(27K) and 450 BeadChip (450K). As implied in its name, 27K covers about 27000 CpG sites, mostly enriched in CpG islands, while 450K covers about 480000 CpG sites encompassing diverse genomic regions selected from previous studies for tDMR, cDMR, rDMR, non-CpG methylated sites, and miRNA promoter regions other than CpG islands.

In this section, we will focus on 450K, the method used in this thesis. The 450K array utilizes bisulfite conversion and genotyping of the C/T polymorphism method to detect methylation at a CpG site quantitatively. Methylated cytosine remains as cytosine, but unmethylated cytosine is converted to uracil after bisulfite conversion (Clark et al., 1994;

Frommer et al., 1992). Note that this method does not distinguish 5mC from 5hmC, because 5hmC remains as cytosine after bisulfite conversion (Huang et al., 2010). For the 450K array, DNA sample is treated with bisulfite, then amplified by PCR. The bisulfite treated and amplified DNA is hybridized on the 450K array that returns the measurement of methylation level at CpG sites on a probe on the array (Dedeurwaerder et al., 2011).

Sequencing-based (Bisulfite pyrosequencing and whole-genome bisulfite sequencing

(WGBS))

Two different levels of sequencing technology are widely used to measure DNA methylation: bisulfite pyrosequencing for a small genomic regions and WGBS for

9 genome-wide level. Bisulfite pyrosequencing is based on bisulfite conversion. As

explained above, methylated cytosine is not affected by bisulfite treatment, while

unmethylated cytosine is converted to uracil, and eventually thymidine after PCR

amplification. Step-wise incorporation of deoxynucleotide triphosphates (dNTPs) during

sequence extension releases pyrophosphates, which are converted to ATP by

sulphurylase. Then, luciferase converts luciferin to oxyluciferin using ATP and produces

light, which will be detected by pyrosequencing machine. The intensity of the released

light is proportional to the amount of the nucleotides incorporated at a single base site.

The C to T ratio can be quantitatively measured by the amount of dCTP or dTTP

incorporation at a cytosine of a CpG site which can be inferred from the intensity of light

released (Bassil et al., 2013). Bisulfite pyrosequencing has been used to detect quantitative methylation and individual CpG sites or to validate results from array-based methods (Doi et al., 2009; Herb et al., 2012; Ji et al., 2010; Kim et al., 2010).

Next-generation sequencing technology has allowed the development of WGBS, enabling researchers to investigate quantitative DNA methylation level at CpG sites genome-wide. WGBS uses bisulfite conversion as in bisulfite pyrosequencing, but the bisulfite converted DNA undergoes next-generation sequencing instead of region specific amplification. The high throughput sequencing data returns read numbers of cytosine versus thymidine, therefore yielding a quantitative measure of DNA methylation at all

CpG sites in the genome (Laird, 2010; Lister et al., 2009). Several statistical methods and software packages have been developed to analyze the WGBS data: BSmooth, Bismark, and so on. Among these packages, BSmooth offers relatively accurate measurement of

DNA methylation at individual CpG sites from low coverage WGBS data by using

10 smoothing algorithm (Hansen et al., 2012). WGBS enables researchers to discover not only DMRs, but also non-CG methylation in stem cells and large genomic DNA methylation changes in cancer and sun exposed skin in elderly people (Hansen et al.,

2011; Lister et al., 2009; Vandiver et al., 2015).

2. Hematopoiesis

Overview

Hematopoiesis is one of the best studied, but complicated developmental systems.

It consists of a hierarchical process initiated by hematopoietic stem cells (HSCs) that differentiate into other hematopoietic progenitor cells and eventually produce all the mature differentiated blood lineages (Chao et al., 2008; Doulatov et al., 2012).

Experimental investigation of the hematopoietic system in the mouse was pioneered by

Till and McCulloch who identified a small subset of cells from the mouse bone marrow which could self-renew and form myeloerythroid colonies (Becker et al., 1963; Till and

Mc, 1961). These studies facilitated other investigators to develop assays such as the in vitro clonal assay combined with fluorescence-activated cell sorting (FACS), to identify and characterize hematopoietic stem and progenitor populations in mouse and human

(Chao et al., 2008; Doulatov et al., 2012). The classical model of hematopoiesis demonstrates that fully differentiated blood cells constitute two major lineages: myeloid and lymphoid. Myeloid lineage cells include granulocytes, monocytes, erythrocytes, and megakaryocytes that give rise to platelets. Lymphoid lineage cells include T, B, and natural killer (NK) cells, involved in immune responses (Doulatov et al., 2012).

11

Human hematopoiesis

Identification of stem and progenitor cells (HSPCs) has promoted our

understanding of human hematopoiesis. Multiple different cell surface markers were

identified and used for isolation of HSPCs. For example, CD34 is a well known marker

for the HSPCs that posses regenerative potential (Civin et al., 1984; DiGiusto et al., 1994;

Krause et al., 1996). Among these CD34+ HSPCs, additional markers such as CD90,

CD38, CD45RA, CD123, and CD10 enable researchers to isolate different components of the hierarchy. Several studies have demonstrated that HSC resides in Lin-CD34+CD38-

CD90+ population (Chao et al., 2008). Further investigation to identify downstream progenitors of HSC has established that multipotent progenitor (MPP) cells are contained in Lin-CD34+CD38-CD90-CD45RA- (Majeti et al., 2007). After HSC gives rise to

MPPs, MPPs further differentiate into progenitors for myeloid or lymphoid lineages. In myeloid lineage differentiation, the common myeloid progenitors (CMPs) develop into either GMPs or megakaryocyte/erythrocyte progenitors (MEPs). Interleukin-3 receptor

alpha chain (CD123) and CD45RA distinguish CMP, GMP and MEP: CMPs reside in

Lin-CD34+CD38+CD123+CD45RA-, GMPs in Lin-CD34+CD38+CD123+CD45RA+,

and MEPs in Lin-CD34+CD38+CD123-CD45RA- (Chao et al., 2008). Lymphoid lineage

differentiation is more complicated, since L-MPP is able to generate lymphoid lineage

cells, as well as monocytes, macrophages and dendritic cells by differentiating into GMP.

This L-MPP population is contained in Lin-CD34+CD38-CD90-CD45RA+ (Doulatov et

al., 2010; Goardon et al., 2011). This thesis will demonstrate DNA methylation

12 differences among human HSPCs and compare epigenetic plasticity between human and

mouse hematopoietic development.

DNA methylation in normal hematopoiesis

Two studies have investigated genome-wide DNA methylation in mouse

hematopoietic development using array or sequencing based methods (Bock et al., 2012;

Ji et al., 2010). Ji et al. performed CHARM examining 4.6 million CpG sites throughout

the genome for MPPs, common lymphoid progenitors (CLPs), CMPs, GMPs, and

thymocyte progenitors (DN1, DN2, DN3). Global methylation changes were involved in

fate decision at the myeloid or lymphoid commitment stage: decreased methylation for

myeloid and gain of methylation for lymphoid commitment. This first comprehensive

methylome map of hematopoietic progenitor cells in murine hematopoiesis identified

potential novel regulators such as Arl4c and Jdp2, as well as previously known transcription factor for hematopoietic differentiation, Meis1. This study demonstrated

DNA methylation is a core mechanism for hematopoietic development, and epigenetic plasticity regulates lineage commitment (Ji et al., 2010). In another study, Bock et al. investigated genome-wide DNA methylation of HSCs, MPP1s, MPP2s, CMPs, CLPs,

GMPs, MEPs, CD4-T cells, CD8-T cells, B cells, erythrocytes, granulocytes, monocytes using reduced representation bisulfite sequencing (RRBS). They observed similar pattern of epigenetic plasticity in cell fate decision: hypermethylation in CLPs and hypomethylation in CMPs. DNA methylation has played a crucial role for silencing of genes involved in myeloid differentiation in lymphoid lineage cells and vice versa. For

13 example, promoters of key transcription factor (TF) for myeloid differentiation such as

Tal1, or binding sites of significant myeloid TFs such as Gata1 were highly methylated in

CLPs compared to CMPs. This study has demonstrated that the information from the

combination of DNA methylation and gene expression data accurately inferred cellular

identity of different blood cells, underscoring the value of DNA methylation in

hematopoietic development (Bock et al., 2012).

For human hematopoiesis, a recent study provided genome-wide DNA methylation profile for HSPCs. This study used a nano HpaII-tiny-fragment-enrichment- by-ligation-mediated-PCR (nanoHELP) assay to investigate DNA methylation of long- term HSCs (LT-HSCs), short-term HSCs (ST-HSCs), CMPs, and MEPs. Loss of methylation has been observed when ST-HSC differentiated into CMPs, and methylation changes has been correlated with gene expression at this transition, while other

commitments such as CMP to MEP transition did not show statistically significant

correlation between DNA methylation and gene expression. These HSC commitment-

associated methylation patterns were able to predict overall patient survival in three

independent AML patient cohorts, indicating the importance of epigenetic regulation for

normal hematopoietic development (Bartholdy et al., 2014). This thesis will compare our

results of comprehensive methylome map of human hematopoiesis to the mouse studies

and the human study.

3. AML and LSC

AML

14 AML is a genetically heterogeneous cancer of myeloid lineage blood cells, and characterized by an accumulation of immature myeloid lineage blood cells in bone marrow. Since AML is caused by diverse pathogenic mechanisms, it is very important to identify subgroups based on the different features of the disease such as morphology and genetic alterations (Lowenberg et al., 1999). The French-American-British (FAB) classification is the most common method to differentiate the heterogeneous disease based on the morphology of leukemic blasts, indicating the degree of differentiation

(Table 1.1) (Bennett et al., 1976, 1985). Specific cytogenetic abnormalities such as translocations and chromosome rearrangements correlate with particular FAB subtypes

(Table 1.1). These cytogenetic lesions have been used in prognosis to predict clinical outcomes and relapse rate (Table 1.2) (Byrd et al., 2002; Grimwade et al., 1998; Slovak et al., 2000). In addition to cytogenetic abnormalities, other genetic mutations play an important role in leukemogenesis, as about a half of AML cases do not harbor a cytogenetic lesion (Dohner, 2007; Lowenberg et al., 1999). The Cancer Genome Atlas

(TCGA) has provided a public resource of genomic map of over 200 AML patients by performing either whole-genome sequencing or whole-exome sequencing. This comprehensive sequencing for a large cohort of AML patients revealed the mutational landscape of AML. Interestingly, AML genomes have relatively fewer genetic mutations compared to other solid tumors, 13 mutations on average, with only 5 in genes recurrently mutated in AML. Genetic alterations were classified to nine different categories based on their biological functions of genes harboring the alterations: TF fusions (18% of cases), NPM1 mutation (27%), tumor suppressors (16%), DNA methylation enzymes (44%), activated signaling genes (59%), myeloid TFs (22%),

15 chromatin modifiers (30%), cohesion-complex genes (13%) and spliceosome-complex

genes (14%). TF-fusion includes PML-RARA, MYH11-CBFB, RUNX1-RUNX1T1, and

PICALM-MLLT10; TP53, WT1, and PHF6 for tumor suppressors; DNMT3A, DNMT3B,

DNMT1, TET1, TET2, IDH1, and IDH2 for DNA methylation enzymes; FLT3, KIT,

KRAS/NRAS, PTPs, and other Tyr or Ser-Thr kinases; RUNX1, CEBPA, and other

myeloid TFs for myeloid TFs; MLL-X fusions, MLL-PTD, NUP98-NSD1, ASXL1,

EZH2, KDM6A, and other modifiers for chromatin modifying enzymes. This study has

suggested common mutations such as DNMT3A, NPM1, CEBPA, IDH1/2 and RUNX1,

which were mutually exclusive to TF-fusion might be involved in the initiation of AML

(Cancer Genome Atlas Research, 2013).

LSC model

LSC model postulates that AML is organized as a hierarchy like normal

hematopoiesis, in which LSCs give rise to leukemic blast cells like HSCs give rise to normal progenitors and differentiated cells. In the 1990s, Dick’s group identified that a

small subset of CD34+CD38- cells were uniquely able to transplant AML into immune

deficient mice (Bonnet and Dick, 1997; Lapidot et al., 1994). These observations lead to

the hypothesis that LSCs possess increased self-renewal capacity, which enables LSCs to

maintain and propagate the disease by generating bulk cancer cells (Kreso and Dick,

2014). Later, improved xenotransplantation models have revealed that LSC activity can

be identified in other subpopulations as well such as CD34+CD38+ or CD34- (Eppert et

al., 2011; Goardon et al., 2011; Martelli et al., 2010; Sarry et al., 2011; Taussig et al.,

16 2008; Taussig et al., 2010). Recently, surface makers other than CD34 and CD38 were identified to enrich for LSCs among heterogeneous cells. C-type lectin-like molecule-1

(CLL-1) was expressed in one third of CD34+CD38- compartment of 29 AML patients

(van Rhenen et al., 2007). CD96 was highly expressed in CD34+CD38- compartment of about two thirds of AML patients (Hosen et al., 2007). T-cell Ig mucin-3 (TIM3), was elevated in LSC fraction compared to normal HSCs (Jan et al., 2011; Kikushige et al.,

2010). CD47 was highly expressed in LSC, and expression of it protected LSCs from being phagocytosed by macrophages (Majeti et al., 2009b). CD25 and CD32 were also identified as novel marker for LSCs (Saito et al., 2010). Even though, many studies have reported a variety of surface markers for LSC, heterogeneous expression of these markers in patients suggests a complex immunophenotype that cannot be applied universally in

AML.

Recently, several studies have investigated genome-wide gene expression profiles of LSCs compared to HSCs or leukemia progenitor cells (LPCs) (de Jonge et al., 2011;

Eppert et al., 2011; Gentles et al., 2010; Majeti et al., 2009a). Majeti et al. performed the first genome-wide gene expression analysis of LSCs compared with HSCs. They identified 3005 differentially expressed genes, enriched in pathways such as Wnt signaling, MAP kinase signaling, adherence junction, ribosome, and T cell receptor signaling (Majeti et al., 2009a). Gentles et al. identified 52 genes which distinguished

CD34+CD38- LSCs from CD34+CD38+ LPCs through genome-wide gene expression analysis. This study showed 52 genes which were associated with overall, event-free, and relapse-free survival and with therapeutic response (Gentles et al., 2010). De Jonge et al. compared CD34+ fractions with the CD34- subfraction of AML patients and CD34+

17 normal progenitor compartment, and found that the top 50 CD34+ specific genes were

able to predict overall survival of AML patients (de Jonge et al., 2011). These studies

used the cell surface markers CD34 and CD38 to define LSC compartment, following the traditional LSC model. Eppert et al. have defined LSCs functionally using a xenograft assay. They sorted AML cells into 4 different fractions based on CD34 and CD38 expression first, then performed a xenograft assay on these fractions from 16 AML patients. They compared the gene expression profile of the functionally validated LSC to non-LSC or HSCs. The analysis demonstrated that LSC and HSC share a core transcriptional program, indicating the commonality between the two populations would be derived from ‘stemness’ property. The genes related to the stemness were associated with clinical outcome (Eppert et al., 2011).

Cell of origin

A number of both mouse and human studies have investigated the cell of origin in

AML. Mouse studies have typically utilized retroviral oncogene transduction or knock-in models to explore this question and have generally led to the conclusion that committed progenitors, in particular CMP and/or GMP, serve as the cell of origin for most AML models. In one study of MN1-induced AML, retroviral transduction of single CMP, but not GMP or HSC, resulted in the development of AML, indicating tight restriction of transformation by this oncogene (Heuser et al., 2011). In a second study using a mouse model of MLL-AF9 AML, the cell of origin influenced biological properties such as gene expression, epigenetics, and drug responses (Krivtsov et al., 2013). Both of these studies

18 highlight the significance of this question for leukemogenesis and potential therapies. In

contrast to mouse models, inferring the cell of origin in human leukemia is only possible

based on features of the disease. Studies investigating the cell of origin of human AML

using surface immunophenotype and gene expression originally suggested AML LSC

arise from HSC (Kreso and Dick, 2014). However, a more recent study that compared

genome-wide gene expression and surface markers of LSCs to those of normal HSPCs

suggests that LSCs arise from more committed progenitors, including L-MPP and GMP

(Goardon et al., 2011). In blast-crisis chronic myeloid leukemia (CML), GMPs with activation of beta-catenin from patients showed increased self-renewal and leukemogenic potential (Jamieson et al., 2004). Notably, three studies have recently reported that leukemogenic mutations existed in HSC, called ‘pre-leukemic HSC’ that underwent further clonal evolution to give rise to AML LSC. These studies have demonstrated a hierarchy among genetic mutations during clonal evolution of the pre-leukemic HSCs.

For example, mutations in epigenome modifying enzymes such as DNMT3A, IDH1/2 have occurred earlier than mutations of genes involved in activated signaling such as

FLT3 and NPM1 (Corces-Zimmerman et al., 2014; Jan et al., 2012; Shlush et al., 2014).

These two studies suggest that the pre-leukemic HSCs which harbor the early occurring genetic mutations is a cell of origin in AML, which may lead to disease relapse after remission.

4. DNA methylation in AML

Somatic mutations in epigenome modifying enzymes in AML

19 Dysregulation of the epigenome is a common feature in AML, as indicated by the

recent discoveries that a number of epigenome modifying genes are mutated in AML.

These genes include several involved in the regulation of DNA methylation such as

IDH1/2, DNMT3A, and TET2, and modulation of chromatin modifications such as

ASXL1, EZH2, and others (Abdel-Wahab et al., 2012; Cancer Genome Atlas Research,

2013; Ley et al., 2010; Yan et al., 2011). Beyond somatic mutations in these epigenome

modifying factors, characterization of DNA methylation in bulk AML cells has revealed great heterogeneity among patient cases. Figueroa et al. examined ~350 AML patient samples using HpaII tiny fragment enrichment by ligation-mediated PCR (HELP) interrogating ~14000 unique gene loci. This study identified 16 distinct clusters among the patients based on DNA methylation profile, which some of the clusters were associated with particular genetic aberrations such as mutations in CEBPA, NPM1,

AML1-ETO, CBFb-MYH11, and PML-RARA (Figueroa et al., 2010b). This study showed the first epigenetically distinct subtypes in AML, associated with genetic alterations. However, this study was done before genetic mutations in epigenetic modifiers were identified. The mutations in epigenome modifying enzymes induce aberrant methylation in the AML cells. In particular, AML with IDH1 or IDH2 mutations was associated with globally increased DNA methylation (Cancer Genome Atlas

Research, 2013; Figueroa et al., 2010a). MLL fusion or mutations in NPM1, DNMT3A, or FLT3 were associated with decreased DNA methylation. It is interesting to observe that about a half of AML patients (44%) harbored a genetic mutation in DNA methylation enzymes, suggesting a critical role of this process in leukemogenesis (Cancer

Genome Atlas Research, 2013). Recent studies have investigated how genetic mutations

20 in DNA methylation enzymes play an important role in leukemic transformation. Challen

et al. has shown loss of function of DNMT3A disturbed HSC differentiation over serial transplantation, while the LT-HSC compartment expanded (Challen et al., 2012).

DNMT3A mutations have been associated with overexpression of HSC specific genes, such as HOXA and HOXB genes in AML patients (Yan et al., 2011). As mentioned

previously, TET enzymes are known to be involved in DNA demethylation by producing

5hmC intermediates. It has been reported that bone marrow samples from patients with

myeloid malignancies with TET2 mutations showed decreased 5hmC with

hypomethylation compared to normal controls, and the interruption of TET2 function in

mouse model has displayed myeloid-skewed differentiation of HSCs. (Ko et al., 2010;

Moran-Crusio et al., 2011; Quivoron et al., 2011). TET2 mutation has been implicated in

clonal hematopoiesis in elderly individuals (Busque et al., 2012; Genovese et al., 2014;

Jaiswal et al., 2014; Xie et al., 2014). Several studies identified recurrent mutations in

IDH1 and IDH2 in AML patients and their function in leukemogenesis (Figueroa et al.,

2010a; Marcucci et al., 2010; Mardis et al., 2009; Ward et al., 2010). IDH1/IDH2

mutations displayed a neomorphic enzymatic activity generating 2-hydroxyglutarate (2-

HG) from α-ketoglutarate (Dang et al., 2009; Ward et al., 2010). 2-HG, oncometabolite,

has been shown to inhibit TET enzyme activity, and induce promoter hypermethylation

(Figueroa et al., 2010a; Shih et al., 2012). All the studies have demonstrated the

significant role of the DNA methylation enzymes in regulation of epigenome in

hematopoietic differentiation and disease development.

Genetic mutations in chromatin modifying enzymes play a significant role in

leukemogenesis. For example, loss of function mutations of ASXL1 causes a genome-

21 wide loss of H3K27me3 and collaborates with oncogenes to promote leukemogenesis

(Abdel-Wahab et al., 2012). The role of EZH2 mutation in myeloid malignancies is complicated, as both loss of function and gain of function mutations in EZH2, a histone lysine methyltransferase and a member of a PRC2 complex have been implicated in leukemogenesis (Ernst et al., 2010; Lund et al., 2014)

Clinical implication of epigenetics in AML patients

Mutations in DNMT3A, IDH1, IDH2, and TET2 have been linked to clinical prognosis such as risk stratification and therapeutic responses in AML patients (Shih et al., 2012). DNMT3A mutation has been associated with adverse overall survival in intermediate-risk group patients (Ley et al., 2010; Patel et al., 2012). TET2 mutation has shown adverse overall survival in AML patients with intermediate-risk, while IDH1 or

IDH2 mutation, which frequently co-occurred with NPM1 mutations, has shown favorable clinical outcome (Patel et al., 2012).

In addition to the association of genetic mutation in DNA methylation enzymes with clinical outcome, DNA methylation itself has been indicated as a prognostic marker.

Figueroa et al. showed an association of distinct methylation clusters of AML patients to overall survival and demonstrated that 15 genes with aberrant DNA methylation could predict overall survival of AML patients (Figueroa et al., 2010b). Furthermore, quantitative DNA methylation has successfully predicted clinical outcome in AML patients (Bullinger et al., 2010). Deneberg et al. has reported that hypermethylation at polycomb group (PcG) target genes was associated with favorable clinical outcome in cytogenetically normal AML (CN-AML) (Deneberg et al., 2011).

22 Besides the prognostic power of DNA methylation, the alteration of DNA methylome of myeloid malignancies has been a target of therapy, as it is reversible.

Azacitidine and decitabine, DNMT inhibitors, are widely used to improve clinical outcome in AML (Estey, 2013). Recent studies have reported that drugs regulating chromatin modification could be an effective therapy for AML patients. For example, suppression of bromodomain-containing 4 (BRD4), which recognizes acetylated lysine on histone, by the small molecule inhibitor, JQ1, demonstrated anti-leukemic effects

(Valent and Zuber, 2014; Zuber et al., 2011). Small molecule inhibitors of DOT1L, a telomeric silencing 1-like histone 3 lysine 79 (H3K79) methyltransferase, have shown therapeutic effects against MLL-fusion AML (Daigle et al., 2013; Daigle et al., 2011).

The epigenetic therapy is an active area of research and clinical trials for AML patients.

23 Neuron Brain

Pluripotent Hepatocyte Liver Stem cell

Blastocyst

Tubular epithelial cell Kidney

Genomic DNA

Figure 1.1. The role of epigenetics in multicellular organism.

24 Table 1.1. French-American-British (FAB) classification.

Type Name Cytogenetics

M0 Undifferentiated acute myeloblastic leukemia

M1 Acute myeloblastic leukemia with minimal maturation

M2 Acute myeloblastic leukemia with maturation t(8;21)(q22;q22), t(6;9)

M3 Acute promyelocytic leukemia (APL) t(15;17)

M4 Acute myelomonocytic leukemia inv(16)(p13q22), del(16q)

M4eo Acute myelomonocytic leukemia with eosinophilia inv(16), t(16;16)

M5 Acute monocytic leukemia del (11q), t(9;11), t(11;19)

M6 Acute erythroid leuekmia

M7 Acute megakaryoblastic leukemia t(1;22)

25 Table 1.2. Cytogenetics and prognosis.

Risk groups Cytogenetics 5-year survival Relapse rate

Good t(8;21), t(15;17), inv(16) 70% 33%

Intermediate Normal, +8, +21, +22, del(7q), Abnormal 11q23 48% 50%

Poor -5, -7, del(5q), Abnormal 3q, Complex cytogenetics 15% 78%

26

Chapter2

Epigenetic signature of leukemia stem cell

27 This work is an ongoing project of the Feinberg Lab and Johns Hopkins. All publication

rights are reserved for these institutions and the presentation of this work here does not

preclude future publication elsewhere.

Summary

AML is a hematopoietic malignancy, composed of a hierarchy that LSCs give rise

into Blast cells. Since LSC is clonally related to Blasts, we hypothesized that particular epigenetic features would endow the distinct capacity of LSC, which can initiate and propagate the disease. Here, we first demonstrated epigenetic differences between LSCs and Blasts by performing genome-wide methylation analysis. We identified 84 DMRs in

70 genes, so called LSC epigenetic signature, which have shown differential methylation and expression. We found that HOXA cluster genes were enriched in LSC epigenetic signature, suggesting a critical role of these genes to confer the unique ability of LSC.

The LSC epigenetic signature was partially dependent on genetic mutation in upstream regulators and epigenome modifying enzymes, yet about a half of it was independent of genetic mutation. The LSC epigenetic signature showed prognostic power in both DNA methylation and gene expression data sets, independent of previously known clinical factors such as age. These results provide the first evidence for epigenetic variation between LSC and their blast progeny in AML, and moreover, demonstrate that DMRs define prognostic subgroups of AML.

Results

28 AML LSC and Blasts Exhibit Epigenetic Differences That Define an LSC Epigenetic

Signature

To formally investigate epigenetic differences between LSC and blast progeny, we sought to identify DMRs between functionally-defined AML LSC-enriched populations and their downstream non-engrafting blasts from a cohort of 15 primary patient samples. We obtained samples from 15 AML patients (Tables 2.1 and 2.2) and isolated subpopulations based on the expression of CD34 and CD38 including: Lin-

CD34+CD38-, Lin-CD34+CD38+, and Lin-CD34- (Figure 2.1). We then performed comprehensive genome-scale DNA methylation analysis using the Illumina Infinium

HumanMethylation450 bead chip array. While AML LSC were originally described to be exclusively contained in the CD34+CD38- subpopulation, recent reports have indicated that leukemia-initiating cells can also be detected in multiple compartments including both the CD34+CD38+ and CD34- subpopulations, although usually at lower frequencies

(Eppert et al., 2011; Goardon et al., 2011; Sarry et al., 2011).

In order to identify LSC and blast populations, we conducted xenotransplantation assays on all three CD34/CD38 subpopulations from each of the 15 AML cases (Table

2.3). Similar to other reports, leukemic engraftment was observed from at least one subpopulation in 10 out of 15 AML patients. As expected, LSC activity dramatically decreased following the immunophenotypic hierarchy with 64.3% of CD34+CD38-,

46.7% of CD34+CD38+, and 26.7% of CD34- subpopulations engrafting in vivo (Table

2.3). To identify epigenetic markers of functional LSC, we performed DMR analysis between the 20 LSC-containing (engrafting) and 24 blast-containing (non-engrafting) fractions (hereafter termed “LSC” and “Blast”). The analysis identified 3030 DMRs, of

29 which 91.4% were hypomethylated in LSC (Table 2.4, see Appendix1 and Table 2.5).

These DMRs were further classified according to their global genomic location

including: islands (regions with a GC content greater than 50% and an observed/expected

CpG ratio of more than 0.6), shores (regions within 2kb of an island), shelves (regions 2

to 4kb away from an island), and open sea (isolated CpG sites in the genome without a

specific designation). These DMRs were nearly evenly distributed in CpG islands

(27.8%) and open seas (29%), (Table 2.5). In addition, the DMRs strongly correlated

with gene expression at CpG islands and open seas, whereas most hypomethylated DMRs

in the engrafting populations were associated with transcriptional up-regulation of

associated genes (Figure 2.2).

We next sought to integrate DNA methylation with gene expression analysis to

identify an LSC epigenetic signature by extracting genes which passed a DMR p value

<0.01 cutoff and exhibited >0.5 log2 fold gene expression change between the LSC and

Blast populations, with an inverse relationship between gene expression and DNA

methylation within 2kb of the transcriptional start site (TSS). We excluded gene body

DMRs, as there was no statistically significant positive correlation in AML or normal

hematopoiesis comparisons (Figure 2.3). We applied a minimum absolute value log2 0.5 fold gene expression cutoff, similarly to our previous LSC gene expression signature using the same microarray platform (Gentles et al., 2010). With these parameters, we identified 84 regions of 70 unique genes exhibiting differential methylation and gene

expression in LSC compared to Blasts (Table 2.6).

We compared our LSC epigenetic signature to the LSC gene expression

signatures from previous studies (Eppert et al., 2011; Gentles et al., 2010). Only six out

30 of 70 genes were found in these earlier studies, suggesting most of the genes identified

here comprise a novel signature for LSC defined first by DMR analysis and refined by

gene expression differences. One gene in this signature, REC8, which encodes a kleisin

family protein that is associated with the cohesin complex, was hypomethylated and

transcriptionally up-regulated in LSC (Figure 2.4a and Table 2.6). Notably, mutations of

components of the cohesin complex have been identified in AML and other tumor types

(Losada, 2014; Thol et al., 2014). We speculate that hypomethylation and increased

expression of REC8 in LSC might be related to cohesin complex activity in LSC. We also

identified HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 in the LSC epigenetic signature (Figure 2.4b-d and Table 2.6). These HOXA cluster genes were hypomethylated and highly expressed in LSC (Table 2.6). Notably, HOXA9 showed hypomethylation and

increased expression in LSC (Figure 2.4b and Table 2.6), and aberrant expression of

HOXA9 is known to be involved in increased proliferation of HSPCs and leukemogenesis,

suggesting a critical role in LSC activity (Chung et al., 2006; Lehnertz et al., 2014;

Takeda et al., 2006; Thorsteinsdottir et al., 2002).

Because the MLL subtype is itself associated with changes in expression of

members of the HOXA gene cluster (Drabkin et al., 2002; Milne et al., 2002), we

performed a second DMR analysis excluding the 5 LSC populations from the 2 MLL

patients in our cohort. We observed substantial overlap between the sets of DMRs

without MLL cases and with all samples. For the key LSC epigenetic signature, 81 of 84

DMRs, including the HOXA genes, were present after removal of the MLL cases (Table

2.7). Considering all DMRs with p < 0.01 (not just the LSC signature), there was 77%

31 overlap (Table 2.7). Thus, the presence of the MLL subtype was not a confounding

variable in defining the LSC epigenetic signature.

The LSC Epigenetic Signature is Partially Dependent on Underlying Somatic

Mutations

In order to identify important pathways and upstream regulators of LSC activity,

we utilized Ingenuity Pathway Analysis (IPA). The most significantly enriched pathway

was fatty acid α oxidation (Table 2.8), and inhibitors of this pathway have been

previously shown to induce apoptosis of leukemia cells (Samudio et al., 2010). Ingenuity

upstream regulator analysis identified NPM1, ASXL1, and KAT6A as the most significant upstream regulators of the LSC epigenetic signature genes, primarily through regulation of HOXA genes including HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 (Table 2.9).

Significantly, all three of these upstream regulators have been found to be mutated in

AML and likely serve as driver genes (Cancer Genome Atlas Research, 2013). In particular, mutations in ASXL1 and NPM1 have been shown to cooperate with HOX

genes to initiate leukemia by enhancing self-renewal and proliferation of hematopoietic

progenitors (Abdel-Wahab et al., 2012; Vassiliou et al., 2011). Consistent with this, we

observed that NPM1 mutation was associated with decreased methylation and increased expression of HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 compared to NPM1 wild-type samples in the TCGA cohort (Figure 2.5).

We then sought to investigate the LSC epigenetic signature for its association

with AML mutations in the TCGA cohort (Figure 2.6). The TCGA cohort consists of 200

AML patient samples with associated DNA methylation, gene expression, and full

genotyping from genome/exome sequencing (Cancer Genome Atlas Research, 2013).

32 First, we identified the epigenetic signatures associated with individual AML mutations

by performing DMR analysis between wild-type and mutant patient samples (Figure 2.6).

The mutations tested included epigenome modifying enzymes such as DNMT3A, IDH1/2, and TET1/2, and upstream regulators of our LSC epigenetic signature, NPM1 and ASXL1

(Figure 2.6). KAT6A was not included as there was no patient who had this mutation among the patients investigated on methylation arrays. Next, we examined the overlap between the mutation-associated DMRs and our LSC epigenetic signature (Figure 2.6).

Each LSC epigenetic signature gene was classified into three categories: (1) upstream regulator-associated if differentially methylated in association with any mutation in upstream regulators; (2) epigenome modifying enzyme-associated if differentially methylated in association with any mutation in epigenetic enzymes; or (3) mutation- independent if it was not differentially methylated in association with either upstream regulator or epigenome modifying enzyme (Figure 2.6 and Table 2.10). Of the 84 LSC

DMRs, 28 (33.3%) and 27 (32.1%) were associated with upstream regulator or epigenome modifying enzyme mutations, respectively (Figure 2.6 and Table 2.10).

However, 40 DMRs (47.6%) including HOXA7 and HOXA9 were mutation-independent targets (Figure 2.6 and Table 2.10). It should be noted that some of the LSC differentially methylated genes, including HOXA7 and HOXA9, have multiple DMRs regulated by different mechanisms. For example, HOXA7 has 4 DMRs in the LSC epigenetic

signature; one associated with mutation in NPM1, two associated with mutation in

DNMT3A, TET1, and NPM1, and one mutation-independent (Table 2.10). Therefore, we

annotated each DMR in those genes differently with DMR numbering such as

HOXA9/DMR1 (Figure 2.6 and Table 2.10). A small subset (11 signatures) of upstream

33 regulator and epigenetic enzyme associated LSC epigenetic signatures overlapped,

including REC8, HOXA6, and HOXA7 (Figure 2.6 and Table 2.10). This analysis showed that all the HOXA genes are epigenetically regulated by at least one upstream regulator, and HOXA6, HOXA7/DMR2, and HOXA7/DMR3 are common targets of both upstream regulators and epigenetic enzymes (Figure 2.6 and Table 2.10), and all of these changes involved DNA hypomethylation. In addition, hypomethylation of HOXA7/DMR1 occurred independently of mutations (Figure 2.6 and Table 2.10). Together, these results suggest that overexpression of HOXA genes mediated by DNA hypomethylation is a core

mechanism for LSC activity.

The LSC Epigenetic Signature is Associated with Overall Survival in Human AML

We hypothesized that if the LSC epigenetic signature reflected key drivers of the

functional differences between LSC and Blasts, then this signature should be associated

with clinical outcomes in human AML. First, we tested the association between the LSC

epigenetic signature and overall survival in the DNA methylation data from the TCGA

AML cohort (Cancer Genome Atlas Research, 2013). To assign each TCGA patient to an

LSC-like or Blast-like category, we calculated scores of each TCGA sample based on the

probability of being closer to either LSC or Blasts. A comparable number of samples

were assigned to each category by this method (99 for Blast-like and 93 for LSC-like). In

univariate survival analysis, the LSC-like group showed worse outcome compared to the

Blast-like group (hazard ratio (HR) =2.3, (95% confidence interval (CI) =1.6-3.4);

p=1.07 x10-5) (Figure 2.7a). The LSC-like vs Blast-like stratification remained associated with overall survival in multivariate analysis together with other known prognostic factors such as age (considered as a continuous variable), cytogenetic risk (assessed as

34 high vs low risk and intermediate vs low risk), NPM1, and FLT3 mutations (HR=1.9,

(95% CI= 1.2-2.9); p=0.003; Table 2.11).

Next, we tested the association between expression of LSC epigenetic signature genes and clinical outcome using four different cohorts including TCGA (Cancer

Genome Atlas Research, 2013), a cohort of normal karyotype patients (Dufour et al.,

2010; Metzeler et al., 2008), and two cohorts of mixed karyotype patients (Valk et al.,

2004; Wilson et al., 2006; Wouters et al., 2009). These cohorts consist of a total of 776

AML patients treated on different clinical protocols that also exhibited distinct biological characteristics(Gentles et al., 2010). We observed a strong correlation between the relative expression of LSC epigenetic signature genes and overall survival in the TCGA cohort (correlation=0.49; p=4 x10-13; Figure 2.8). The more highly expressed a gene was in LSC compared to Blasts, the more robust its association with worse overall survival. In all four cohorts, the overall expression level of the signature genes was significantly associated with overall survival, with higher expression associated with worse clinical

(Table 2.12). This association remained significant in multivariate Cox regression including age (continuous), cytogenetic risk, NPM1, and FLT3 mutations (HR=1.7, (95%

CI, 1.0-2.7); p= 0.03; Table 2.11). Similar results were observed for the three other cohorts in univariate and multivariate analyses (Figures 2.7c-e, Tables 2.12 and 2.13).

Finally, we tested if mutations in epigenetic enzymes such as DNMT3A, IDH1/2,

TET2, and ASXL1 affected the prognostic impact of the LSC epigenetic signature in the

TCGA cohort. As described previously, mutation in DNMT3A, but none of the other genes, was associated with patient overall survival (Table 2.14). Multivariate survival analysis including DNMT3A mutation showed that our LSC epigenetic signature

35 remained independently associated with clinical outcome in both the DNA methylation and gene expression data from TCGA, even when incorporating cytogenetic risk group

(Table 2.15), as well as within the intermediate cytogenetic risk group alone (Table 2.16).

Overall, these results demonstrate that the LSC epigenetic signature defined by DNA methylation and gene expression is associated with overall survival in human AML.

Discussion

The cancer stem cell (CSC) model was originally proposed based on observations from human AML in which only subpopulations of leukemia-initiating or LSC possessed engraftment potential (Bonnet and Dick, 1997; Lapidot et al., 1994). According to this model, the LSC give rise to downstream Blasts that lack critical stem cell properties. As

LSC and their non-engrafting Blast progeny are clonally related, a major implication of this leukemia stem cell model is that their functional properties must be due to epigenetic differences. Here, we provide such evidence by characterizing global DNA methylation features of LSC defined by xenotransplantation of AML subpopulations, compared to non-engrafting Blast cells, demonstrating that AML LSC exhibit global hypomethylation compared to non-LSC Blast cells. Integrating DNA methylation and gene expression analysis, we identified 84 regions of 70 genes as the LSC epigenetic signature. 64 of these 70 genes were not reported in previous gene expression studies for LSC (the exceptions being CD34, SH3BP5, RBPMS, LTB, MS4A3, and VNN1) (Eppert et al., 2011;

Gentles et al., 2010). Most of the LSC epigenetic signature was mutation-independent, not associated with mutations in upstream regulators or epigenome-modifying enzymes suggesting that leukemogenesis may converge on these primary epigenetic signatures We

36 also identified some mutation-associated epigenetically dysregulated genes, including

REC8 and HOXA7. Together, these epigenetic signatures represent potential therapeutic targets regardless of the different types of the underlying mutations present in individual

AML cases. Furthermore, the LSC epigenetic signature was prognostic of patient overall survival independently of known survival predictors such as age and cytogenetic abnormalities, emphasizing its functional importance.

Apart from its prognostic effect, the LSC epigenetic signature represents a

molecular target that may improve patient survival and prevent relapse. Recently,

epigenetic therapy with hypomethylating agents azacytidine and decitabine has been

approved for the treatment of AML. Randomized trials demonstrated improved overall

survival compared to chemotherapy, but also indicated limited effect on relapse rate in

high-risk AML (Estey, 2013). Our results indicate that LSC are relatively

hypomethylated compared to Blasts, suggesting that they may be less effectively targeted

by hypomethylating agents, possibly accounting for their limited efficacy in relapse-free

survival. It would be of great interest to see how the LSC epigenetic signature is affected

by these drugs.

More specifically, this LSC epigenetic signature was markedly enriched for

members of the HOXA cluster, suggesting this cluster is a key driver of LSC function.

The HOXA cluster has been implicated as a key regulator of hematopoiesis and myeloid malignancy (Alharbi et al., 2013). In particular, HOXA9 is known to be involved in increased proliferation of HSPC and leukemogenesis (Thorsteinsdottir et al., 2002), even occurring as a fusion oncogene in rare cases (Nakamura et al., 1996). Moreover, increased expression of HOXA9 has been found to be an adverse prognostic factor in

37 AML (Golub et al., 1999). Other HOXA family members including posterior (HOXA7,

HOXA9, and HOXA10) and anterior (HOXA6) members have been implicated in leukemogenesis, as overexpression of these genes in normal mouse HSPC leads to increased self-renewal, transformation, and development of myeloid malignancies (Bach

et al., 2010). The functional LSC epigenetic signature provided here demonstrates that

the HOXA family is a key driver of AML LSC that may function in imparting aberrant

self-renewal.

Materials and Methods

Human Samples

Human acute myeloid leukemia (AML) samples were collected from patient peripheral

blood (PB) or bone marrow (BM) at Stanford hospital, according to an IRB-approved

protocol (22264), and informed consent was obtained from all subjects. PBMC or

BMMC were separated with Ficoll-Paque Plus (Amersham Biosciences, Piscataway, NJ,

Catalog number: 17-1440-03), and cryopreserved in 1 x freezing medium (90%FBS +

10%DMSO). All the AML experiments were conducted with cryopreserved PBMC or

BMMC samples that were thawed and washed in IMDM medium containing 10% FBS.

Flow Cytometry Analysis and Cell Sorting

A battery of antibodies (Abs) was used for staining, analysis and sorting of progenitor

cells from AML patient PBMCs/BMMCs, as well as lineage analysis human

chimerism/engraftment (Table 2.17). Cells were either analyzed or sorted using a FACS

Aria II cytometer (BD Biosciences, Franklin Lakes, NJ). Analysis of flow cytometry raw data was done with FlowJo Software (Treestar, Ashland, OR).

38 Xenotransplantation Assay

NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice (NSG) were obtained from The Jackson

Laboratory (Bar Harbor, ME) and bred in a specific pathogen-free environment per

Stanford Administrative Panel on Laboratory Animal Care Guidelines (Protocol 22264).

Six to eight week-old adult mice were exposed to 200 rads of gamma irradiation at least two hours (up to 24 hours) prior to transplantation. Up to 500 thousand fresh-sorted

AML cell subpopulation were resuspended in 30 µl of Hank’s Balanced Salt Solution

(HBSS) (Gibco Life Technologies, Grand Island, NY) containing 2% FBS, and injected intravenously via the tail vein using a 29-gauge needle. For each cell subpopulation, at least three technical replicates were performed by transplantation of three aliquot of cells into three mice. Around 150 mice in total were used. Neither randomization nor blinding was used for this study.

After eight weeks, mice were euthanized with CO2 according to IRB approved protocol

(22264). BM were isolated using scissors and needle flashing, then underwent hypotonic red cell lysis using ACK (Ammonium-Chloride-Potassium) lysing buffer (Life

Technologies, Grand Island, NY, Catalog# A10492). BMMCs were stained with Ab combinations (Table 2.17) on ice for 30 minutes, and dead cells were excluded by propidium iodide (PI) staining. Human myeloid engraftment (hCD45+CD33+) and lymphoid engraftment (hCD45+CD19+) were analyzed on flow as described before.

Illumina Infinium Human Methylation 450 Bead Array Assay

Genomic DNA from each sample was purified using the MasterPure DNA purification kit (Epicentre) according to the manufacturer’s protocol. The genomic DNA (250-500ng) was treated with sodium bisulfate using the Zymo EZ DNA Methylation Kit (ZYMO

39 Research) as recommended by the manufacturer, with the alternative incubation

conditions for the Illumina Infinium Methylation Assay. Converted DNA was eluted in

11ul of elution buffer. DNA methylation level was measured using Illumina Infinium HD

Methylation Assay (Illumina) according to the manufacturer’s specifications.

Methylation array data are deposited at the Gene Expression Omnibus (GEO) with

accession number GSE63409.

Illumina Infinium Human Methylation 450 Bead Array Analysis

Raw intensity files were obtained using minfi package (Aryee et al., 2014) to calculate methylation ratios (Beta values). The data was normalized using Illumina preprocessing method implemented in minfi. Several quality control measures were applied to remove

arrays with low quality. Control probes were examined on the 450k array to assess several measures including bisulfite conversion, extension, hybridization, specificity and others. Next, median methylated and unmethylated signals were calculated for each arrays; no array was identified for signal values lower than 10.5. For multidimensional scaling analysis, probes containing an annotated SNP (dbSNP137) at the single-base extension or CpG sites were removed (17398 probes removed). Minfi 1.8.9 was used.

Bump hunting method previously described was applied to identify DMRs in 450k array

(Aryee et al., 2014; Jaffe et al., 2012). Beta value of 0.1 (10% of methylation difference) was used as cutoff when finding DMRs. Statistical significance was assigned by permutations testing and the P-value cutoff used for downstream analysis was <0.01 that corresponded to Benjamini-Hochberg adjusted p-value <0.1 (data not shown) unless different cutoff was designated in result part. Bumphunter 1.2.0 was used. Same method

40 was applied to identify DMRs for the second DMR analysis of LSC vs Blast that we removed 5 LSC cases from 2 MLL patients (SU042 and SU046).

Affymetrix Microarray Expression Analysis

Total RNA was extracted from each FACS-sorted cell population using RNeasy® Plus

Mini (QIAGEN, Valencia, CA, Catalog#: 74134) according to the manufacture’s protocol. All RNA samples were quantified with 2100 Bioanalyzer (Agilent

Technologies, Santa Clara, CA), subjected to reverse transcription, two consecutive rounds of linear amplification, and production and fragmentation of biotinylated cRNA.

15µg of cRNA from each sample was hybridized to HG U133 Plus 2.0 microarrays.

Hybridization and scanning were performed according to the manufacture’s instruction

(Affymetrix). This step was performed at the PAN center of Stanford University. Data were normalized by GC robust multi-array average method and analyzed on

R/Bioconductor. SU042 CD34+38+ was removed from further analysis due to low quality. SU001 was excluded, as the samples from this patient were not included on expression array (GEO GSE63270).

Sanger Sequencing to Detect AML Mutations

Genomic DNA was extracted from patient BMMC or PBMC using QIAmp DNA Mini

Kit (QIAGEN, Valencia, CA, Catalog#: 51304) according to the manufacture’s instruction. PCR primers were designed to cover exon 3-11 of TET2, exon 4 of IDH1/2, and exon 7-23 of DNMT3A (Table 2.18). The PCR reaction premix consists of 1x of

OneTaq 2x Master Mix (NEB, Ipswich, MA, Catalog#: M0482L), 0.2µM forward and reverse primers respectively, and 10ng (up to 100ng) genomic DNA as template. The reaction was under the condition of 95ºC initial denaturation for 30 seconds, 45 cycles of

41 extension containing 94ºC for 30 seconds, 56ºC for 1 minute (or as necessary) and 72ºC for 1 minute, and a final extension at 72ºC 5 minutes. The PCR products were concentrated with PCR purification kit (QIAGEN, Valencia, CA, Catalog#: 28106), then submitted to Sequentech (Mountain View, CA) for sequencing of both forward and reverse directions using 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA) according to the manufacturer’s instruction. The sequencing data was analyzed using

Sequencher 5.1 (Gene Codes Corporation, Ann Arbor, MI), and single-nucleotide polymorphism (SNP) was excluded by checking NCBI website before getting the final mutation results.

Survival Analysis

Survival analysis was performed to assess the association of LSC DNA methylation and gene expression signatures with clinical outcome (overall survival) in 4 different cohorts.

For DNA methylation data set (TCGA), patients were separated into two groups; LSC- like and Blast-like based on the methylation profile of each individual. Survival was compared between the two groups using the coxph function in R (survival package 2.37), with significance assessed by log-rank test. For gene expression, the genes in the LSC epigenetic signature were identified in expression datasets for which survival outcomes were available. The first principal component of their expression levels was computed, and patients were stratified as “high” or “low” relative to its median value. Survival differences between the groups were assessed by log-rank test. In multivariate analyses, age was incorporated as a continuous variable, mutations were coded as present/absent

(1/0), and assessment of cytogenetic risk was treated as individual groups and done for

42 intermediate vs low-risk and High vs low risk (Figure 2.9). Analysis was also performed within intermediate risk groups.

Statistical Analysis

To assign cell identity of LSC/Blast to TCGA samples, mean methylation value of each

LSC epigenetic signature (84 DMRs) for LSC/Blast (methylation profile) was retrieved and standard deviation of the mean value for each signature was calculated. Then scores

(probability density values as log value) for each TCGA sample regarding LSC and Blast profile was calculated using dnorm function with the mean and standard deviation calculated in previous step. Maximum value of scores between the ones regarding LSC and Blast methylation profile was chosen, and then cell identity assigned.

Bioinformatics Analysis

QIAGEN’s Ingenuity IPA (Ingenuity® Systems, www.ingenuity.com) was performed for pathway analysis.

43

Supplementary Figure 1. Pre-sort and post-sort FACS analysis of subpopulations from human AML. Top panel: FACS-sorting scheme of three immunophenotypically defined subpopulations from human AML samples. Other panels: Two rounds of post- sort analysis to check the purity of sorting.

44 Figure 2.1. Pre-sort and post-sort FACS analysis of subpopulations from human

AML. Top panel: FACS-sorting scheme of three immunophenotypically defined subpopulations from human AML samples. Other panels: Two rounds of post-sort analysis to check the purity of sorting.

45 Island Shore Differential expression Differential expression

Differential methylation Differential methylation Shelf Open sea Differential expression Differential expression Differential methylation Differential methylation

Supplementary Figure 2. Gene expression inversely correlates with DMRs at CpG island and open sea. Engrafting (LSC) and non-engrafting (Blast) subpopulations from primary AML cases were profiled for DNA methylation and gene expression to identify differentially methylated regions (DMRs) and differentially expressed genes between these two groups DMRs that are located within kb of gene transcriptional start sites (TSSs - black dots) were classified into groups according to their distance relative to a CpG island: island shore shelf and open sea DMRs located further than kb away from TSSs are denoted as black pluses Log ratios of differential expression were plotted against differential methylation (all values are Blast compared to LSC) Wilcoxon rank-sum tests were performed to test the null hypothesis that the expression differences for the hypo- or hypermethylated DMRs within kb of gene TSSs (black dots) showed stronger inverse correlation than the expression differences of the random DMRs that are located further than kb of TSSs (black pluses) Random DMRs were shown in the middles of DNA methylation axis regardless of their methylation differences

46 Figure 2.2. Gene expression inversely correlates with DMRs at CpG island and open sea. Engrafting (LSC) and non-engrafting (blast) subpopulations from primary AML cases were profiled for DNA methylation and gene expression to identify differentially methylated regions (DMRs) and differentially expressed genes between these two groups.

DMRs that are located within 2kb of gene transcriptional start sites (TSSs - black dots) were classified into 4 groups according to their distance relative to a CpG island: island, shore, shelf, and open sea. DMRs located further than 2kb away from TSSs are denoted as black pluses. Log2 ratios of differential expression were plotted against differential

methylation (all values are blast compared to LSC). Wilcoxon rank-sum tests were

performed to test the null hypothesis that the expression differences for the hypo- or

hypermethylated DMRs within 2kb of gene TSSs (black dots) showed stronger inverse

correlation than the expression differences of the random DMRs that are located further

than 2kb of TSSs (black pluses). Random DMRs were shown in the middles of DNA

methylation axis regardless of their methylation differences.

47 a Island Shore Differential expression Differential Differential expression Differential v v v v Differential methylation Differential methylation

Shelf/Open sea Differential expression Differential v v

Differential methylation

48 b HSC vs GMP Island Shore Differential expression Differential Differential expression Differential v v v v Differential methylation Differential methylation

Shelf/Open sea Differential expression Differential v v Differential methylation

HSC vs MEP

Island Shore Differential expression Differential expression

Differential methylation Differential methylation

Shelf/Open sea Differential expression

Differential methylation

49 Figure 2.3. Gene body methylation doesn’t show statistically significant positive

correlation with gene expression. DMRs that are located in gene body (TSS to

transcription end site (TES)) were classified into three groups according to their distance

relative to a CpG island: island, shore, shelf/open sea. Random DMRs that don’t locate in

gene body are denoted as black pluses. Log2 ratios of differential expression were plotted against differential methylation. Wilcoxon rank-sum tests were performed to test the null hypothesis that the expression differences for the hypo- or hypermethylated DMRs located in gene body (black dots) showed stronger positive correlation than the expression differences of the random DMRs that do not locate in gene body (black pluses). (a) LSC vs Blast. All values for DNA methylation and gene expression are from

LSC-Blast. (b) Normal hematopoiesis. HSC vs GMP and HSC vs MEP are shown. All values for DNA methylation and gene expression are from group2 – group1 for group1 vs group2 comparisons.

50 Figure 1

a b REC8 HOXA9

● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● Beta

Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ●● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● Blast● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● Blast ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ●● ●● ● ● ● ● ● ● ●●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ●●●●●●● ● ● ● 0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4LSC 0.6 0.8 1.0 LSC

0.1 location on chr14 0.1 location on chr7

0 0 + + HOXA9 Genes Genes REC8 chr1 24641000 24641400 24641800 24642200 chr7 27204500 27205000 27205500 27206000

c HOXA7 d HOXA10

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● ● ● ● ● Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● Blast● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● Blast ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● LSC LSC ● ● ● 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.1 location on chr7 0.1

0 0 + + HOXA10 Genes Genes chr7 27197500 27198500 27199500 chr7 27213600 27214000 27214400

Figure 2.4. AML LSC and Blasts exhibit epigenetic differences that define an LSC

epigenetic signature. Plots of differentially methylated regions (DMRs) indicating

genomic loci for REC8 (a), HOXA9 (b), HOXA7 (c) and HOXA10 (d), four LSC

epigenetic signature genes that are hypomethylated and upregulated in LSC. Top: level of

CpG methylation (beta) of each sample for the region; Middle: CpG density (curve), CpG

sites (black tick marks), CpG islands (red lines); Bottom: gene annotation.

51 a HOXA5 HOXA6 HOXA7 Methylaion Methylaion Methylaion 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

p=0.0001 p<2x10-16 p<2x10-16

HOXA9 HOXA10

NPM1 Mutant Methylaion Methylaion 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

p=5.4x10 -10 p=1.9x10 -10

52 b HOXA5 HOXA6 HOXA7 10

10

10 8 8

6 6 5

4 4

-16 -16 -16 p<2x10 p=9x10 p<2x10 0 2 2

HOXA9 HOXA10 Expression level NPM1 10.0

10 Mutant 7.5

5 5.0

p<2x10-16 p<2x10-16 2.5

Supplementary Figure 4. NPM1 mutation is associated with decreased methylation and increased expression of HOXA genes. (a) Box plots show methylation level for NPM1 mutants and wild-type samples for DMRs for HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 in the TCGA dataset. t-test assuming unequal variance was performed to look at statistical significance of the association between NPM1 mutation and methylation. DNA methylation of all the HOXA genes was

significantly associated with NPM1 mutation. (b) Box plots show gene expression (Log2 value) for NPM1 mutants and wild-type samples for HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 in the TCGA dataset. t-test assuming unequal variance showed NPM1 mutation highly correlated with increased expression of all the HOXA genes tested.

53 Figure 2.5. NPM1 mutation is associated with decreased methylation and increased expression of HOXA genes. (a) Box plots show methylation level for NPM1 mutants and

wild-type samples for DMRs for HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 in the

TCGA dataset. t-test assuming unequal variance was performed to look at statistical significance of the association between NPM1 mutation and methylation. DNA methylation of all the HOXA genes was significantly associated with NPM1 mutation. (b)

Box plots show gene expression (Log2 value) for NPM1 mutants and wild-type samples for HOXA5, HOXA6, HOXA7, HOXA9, and HOXA10 in the TCGA dataset. t-test

assuming unequal variance showed NPM1 mutation highly correlated with increased expression of all the HOXA genes tested.

54 Upstream regulator Epigenome modifying LSC epigenetic signature Identify upstream regulators enzyme Integrate DNA methylation (p value <0.01) of LSC epigenetic signature Frequently mutated in AML and gene expression analysis (Log2 ratio>0.5) by Ingenuity analysis DNMT3A, IDH1/2 for LSC vs blast NPM1 and ASXL1 and TET1/2 84 DMRs in 71 genes

TCGA Identify DMRs (p value<0.01) associated with mutation in upstream regulators or epigenome modifying enzymes

Look for overlap

LSC epigenetic signature LSC epigenetic signature LSC epigenetic signature NOT associated with either associated with mutation in associated with mutation in

upstream regulators epigenome modifying enzymes HOXA9/DMR1, HOXA9/DMR2 MPO/DMR1, HOXA9/DMR3 SCRN1/DMR1, FAM24B HOXA9/DMR5, CPNE3, MANSC1 HOXA9/DMR4, CD34, HOXA10 FAM169A, MPPED2, HOXA7/DMR2* NDFIP1, PTGER4, HOXA7/DMR1 HOXA7/DMR2*, CPNE8, ROBO3 RHBDF1, REC8*, HOXA6*, DOCK1* EREG, RAB34/DMR1, LTBP3 GSTM5, DDIT4, REC8* , HOXA6* ATP6V0A2, EID3, EPCAM, SH3BP5 RAB34/DMR2, ZFP3, SLITRK5 FAAH, DOCK1*, MPO/DMR2, SCRN1/DMR2, SCRN1/DMR3 MAFF, NAP1L3, ALDH2/DMR1 HOXA5, H1F0, HOXA7/DMR3* HOXA7/DMR3*, LAMB2, CRHBP* ALDH2/DMR2, HOXA9/DMR6 HOXA7/DMR4, CRHBP*, SOCS2 LTB, ERMP1, MPP6, MS4A3* RBPMS, TRIP6, OSBPL3, MOSC2 MS4A3*, VNN1*, LST1* VNN1*, LST1*, CD52, CLIC2* IL12RB2, CYorf15A,PTGS2, SNCA FAM30A, BLNK*, CLIC2*, No name* No name* CHRDL1, GUCY1B3, HOOK1 RPL17, NFIA, MREG, CBR1 TMEM22, C10orf10, BGLAP LPAR6, RXFP1, HDAC9 28 DMRs (33.3%) 27 DMRs (32.1%) 40 DMRs (47.6%) * Genes associated with mutation in both upstream regulators and epigenome modifying enzymes

Figure 2.6. The LSC epigenetic signature is partially dependent on underlying somatic mutations. Shown is a schematic flow chart of mutation association analysis.

We compared our LSC epigenetic signature to the mutation specific DMRs obtained from

TCGA data set. The LSC epigenetic signature was classified into three different groups, and each DMR is shown in this diagram. Note that several genes such as HOXA9 have multiple DMRs and different DMRs in one gene are annotated with DMR number such as HOXA9/DMR1.

55 Figure 3

a b TCGA DNA methylation TCGA Gene expression Blast-like Low score LSC-like High score p=1.1X10-5 p=1.0X10 -5 HR=2.3 HR=2.4 Overall survival Overall survival 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0 20 40 60 80 0 20 40 60 80 Months Months c Metzeler et al. d Wouters et al. e Wilson et al. Gene expression Gene expression Gene expression Low score Low score Low score High score High score High score p=1.0X10-3 p=2.0 X10-7 p=2.4X10-6 HR=1.9 HR=2.3 HR=2.2 Overall survival Overall survival Overall survival 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 0 50 100 150 0 20 40 60 80 100 120 Months Months Months

56 Figure 2.7. The LSC epigenetic signature is associated with overall survival in human AML. (a) TCGA samples were classified as LSC-like or Blast-like based on

DNA methylation alone by generating methylation profiles of the LSC and Blast populations, and then calculating scores of each sample based on the probability of being closer to either LSC or Blast. Kaplan-Meier survival analysis was then applied to these groups as indicated. Statistical significance was determined by the Log-rank test (n=192;

93 LSC-like and 99 Blast-like patients). (b-e) Expression of the LSC epigenetic signature genes was combined to create an LSC score, which was then calculated in AML samples from four independent cohorts including TCGA (n=182; 91 high and 91 low score patients) (b) Metzeler et al (n=163; 81 high and 82 low score patients) (c), Wouters et al

(n=262; 131 high and 131 low score patients) (d), and Wilson et al (n=169, 84 high and

85 low score patients) (e). In each cohort, patients were classified into high and low groups based on the median value. Kaplan-Meier survival analysis was then applied to these groups as indicated. Statistical significance was determined by the Log-rank test.

57 p 0

0

Differential expression(Log)

FigureSupplementary 2.8. The gene Figure expression 5. The of gene the expressionLSC epigenetic of the signature LSC epigenetic highly signature correlates highly correlates with clinical outcome in the TCGA dataset Eah dot represents an with clinicalLSC epigeneti outcome signature in the TCGA gene Survival dataset z-s. Eachore dot was represents plotted against an LSC log epigenetic ratio of differential expression of the LSC epigeneti signature genes in TCGA signature gene. Survival z-score was plotted against log2 ratio of differential expression of the LSC epigenetic signature genes in TCGA.

58 a ## DNA methylation analysis and survival stime <- ifelse(tpd$vital_status=="DECEASED",tpd$days_to_death,tpd$days_to_last_followup) stime[stime%in%c("[Not Available]","[Not Applicable]")]<-NA stime<-as.numeric(stime) event<-tpd$vital_status=="DECEASED" age<-tpd$age_at_initial_pathologic_diagnosis prog<-tpd$acute_myeloid_leukemia_calgb_cytogenetics_risk_category prog[prog=="[Not Available]"]<-NA prog<-factor(prog,levels=c("Favorable","Intermediate/Normal","Poor"),labels=c("F","I","P")) library(survival) summary(coxph(Surv(stime,event)~group+age+prog+Flt3+Npm1) summary(coxph(Surv(stime,event)~group+age+prog+Flt3+Npm1+tpd$dnmt3a) b ## GEP analysis and survival

dmrexp = read.delim("TCGA_AML_newDMR_p001_fc05.eigengenes.pcl", stringsAsFactors=FALSE) amlinfo2 = merge(amlinfo,dmrexp, by="Array") medexp = median(amlinfo2$DMR_p0.01_fc0.5, na.rm=TRUE) amlinfo2$medexp = 1 amlinfo2$medexp[amlinfo2$DMR_p0.01_fc0.5>medexp] = 2 summary(coxph(Surv(OS_Time,OS_Status) ~ DMR_p0.01_fc0.5+NPM+FLT3+dnmt3a+Age+CALGB_cytorisk, data=amlinfo2)) summary(coxph(Surv(OS_Time,OS_Status) ~ DMR_p0.01_fc0.5+NPM+FLT3+Age+CALGB_cytorisk, data=amlinfo2))

Figure 2.9. R script for multivariate survival analysis. (a) Multivariate survival

Suanalysispplementary for DNA Figure methylation 10. R datascript in TCGA. for multivariate The line or survival a variable analysis that show. (a) how we Multivariate survival analysis for DNA methylation data in TCGA. The line or a variable thatreatedt show cytogenetic how we tre atedgroups cytogene is coloredtic gro upsin red. is col (bore) Multivariated in red. (b) Msurvivalultivariate analysis survival for gene analysis for gene expression data in TCGA. The line or a variable that show how we treatedexpression cytogenetic data in groups TCGA. is colored The line in r ed.or a variable that show how we treated cytogenetic

groups is colored in red.

59 Table 2.1. Clinical features of AML patients in this study

Sample % WHO Age Gender 1°/2° D/R Cytogenetics FAB ID CD34+ Classification SU001 59 F 1° R Normal 99 AML-not otherwise specified M2 SU006 51 F 1° D Failed to grow 94 AML-not otherwise specified M1 SU008 64 M 1° D Normal 3 AML-not otherwise specified M1 SU014 59 M 1° D Normal 18 AML-not otherwise specified ND inv(9)(p11q13 AML with multilineage dysplasia SU029 65 F 1° D 8 M2 ) without antecedent MDS SU032 47 M 1° D Normal 68 AML-not otherwise specified M5 SU035 46 M 1° D Failed to grow 98 AML-not otherwise specified M5 SU036 71 F 1° D t(8;21) 47 AML with t(8;21)(q22;q22) ND SU042 61 F 1° D t(10;11) 8 AML with 11q23 (MLL) M5b SU046 53 F 1° D t(6;11) 94 AML with 11q23 (MLL) M5 Complex AML with multilineage dysplasia SU056 56 M 1° D 99 M0 cytogenetics without antecedent MDS SU266 65 M 1° D inv(3) 96 AML with inv(3)(q21q26) ND AML with multilineage dysplasia SU267 58 M 1° D Normal 66 ND without antecedent MDS SU302 59 M 1° D Normal 14 AML-not otherwise specified ND No analyzable SU306 33 F 1° D <1 AML-not otherwise specified M5a metaphases Abbreviations: 1°, primary; 2°, secondary; D, de novo; F, female; M, male; ND, no data; R, relapsed

60 Table 2.2. Genetic mutations identified Patient DNMT3 FLT3 FLT3 TET2 IDH1 IDH2 NPM1 KIT CEBPA ID A ITD TKD SU001 wt wt wt wt wt nd wt nd nd SU006 wt wt wt wt wt nd wt nd nd SU008 wt wt wt wt mut wt wt nd nd SU014 wt R132H wt wt mut nd mut nd nd SU029 1149FS wt wt R882H mut nd mut nd nd SU032 Y1649C wt wt wt wt nd wt nd nd SU035 wt wt wt wt wt nd wt nd nd SU036 wt wt wt wt nd nd wt mut nd SU042 wt wt wt S837* wt nd wt nd nd SU046 wt wt wt wt wt wt wt nd Nd SU056 wt wt wt wt wt wt wt nd wt SU266 E1010D wt wt wt wt wt wt nd wt SU267 wt R132C wt R882H wt wt wt nd wt SU302 wt wt wt R882H wt wt mut wt mut SU306 wt wt R140Q ΔV149 wt mut mut wt wt

Abbreviations: FS, frameshift mutation; wt, wild type; mut, mutant; nd, no data; * stop; Δ, deletion.

Note: Sanger sequencing was performed on TET2 exon 3-11, IDH1, IDH2 exon 4, and DNMT3A exon 3-11. More details are provided in Table 2.18. For all other mutations, data are derived from clinical laboratory testing.

61 Table 2.3. Engraftment of AML subpopulations “CD34+CD38- Patient ID “CD34-” “CD34+CD38+” ” SU001 No No No SU006 No No Yes SU008 No No No SU014 No No No SU029 Yes Yes Yes SU032 No No No SU035 Yes No Yes SU036 No No No SU042 Yes Yes Yes SU046 Yes Yes ND SU056 No Yes Yes SU266 No Yes Yes SU267 No Yes Yes SU302 No Yes Yes SU306 No No Yes 4/15 7/15 9/14 Frequency (26.7%) (46.7%) (64.3%) Note: Yes: engrafted; No: no-engraftment; ND, no data. For SU046, there is no CD34+CD38- cell fraction.

62 Table 2.4. DMRs of LSC vs Blast (See Appendix1)

63 Table 2.5. Summary of LSC vs Blast DMRs

Comparisons Numbers of DMRs* Locations of DMRs relative to CpG islands (%)

(Group1 versus Group2) Group1>Group2 Group1

Blast vs LSC 2769 261 27.8 37.8 5.4 29

* P value cutoff of 0.01 was used to calculate the number of DMRs (see Methods)

64 Table 2.6. LSC epigenetic signature

chr start end diffMethyl island diffexp Gene DMR_TSS_dist

chr17 56356470 56356963 0.200454417 Island -2.771069849 MPO 1304

chr7 27209463 27209582 -0.190449837 Island 2.262250718 HOXA9 -194

chr7 27205200 27205262 -0.205288088 Island 1.750125912 HOXA9 -51

chr7 27203430 27203546 -0.200286397 Island 1.750125912 HOXA9 1603

chr7 27206073 27206907 -0.1623543 Island 1.750125912 HOXA9 -924

chr7 27204052 27204981 -0.150359078 Island 1.750125912 HOXA9 168

chr7 30029717 30029808 -0.232917635 Island 1.682967495 SCRN1 -402

chr1 208083913 208084071 -0.175662022 Island 1.581896622 CD34 612

chr8 87526705 87527257 -0.175070398 Island 1.492929541 CPNE3 -8

chr7 27213984 27214383 -0.149812433 Island 1.460273965 HOXA10 0

chr12 12502846 12502846 -0.262873441 Island 1.309407455 MANSC1 329

chr5 141488047 141488121 -0.196034559 Island 1.163061111 NDFIP1 314

chr10 124638756 124639630 -0.206994825 Island 1.098988409 FAM24B 0

chr5 74162602 74162809 -0.198910627 Island 0.984734984 FAM169A 0

chr11 30605787 30606026 -0.194575258 Island 0.976935267 MPPED2 -226

chr5 40681444 40681444 -0.278087288 Island 0.968009601 PTGER4 -1406

chr7 27195918 27196286 -0.172243476 Island 0.940816384 HOXA7 10

chr7 27198025 27198896 -0.165502698 Island 0.940816384 HOXA7 -1729

65 chr16 122031 122031 -0.238501186 Island 0.9115447 RHBDF1 562 chr4 75230391 75230615 -0.236192175 Island 0.91073247 EREG 244 chr12 39299364 39299726 -0.170515768 Island 0.881992857 CPNE8 0 chr17 27045043 27045302 -0.210388337 Island 0.87035073 RAB34 -97 chr11 65325158 65325249 -0.218616458 Island 0.852913568 LTBP3 139 chr17 27044169 27044685 -0.191399132 Island 0.851673857 RAB34 0 chr11 124747075 124747263 -0.258430121 Island 0.825061807 ROBO3 -338 chr1 110254692 110255096 -0.204217813 Island 0.820222184 GSTM5 0 chr10 74034644 74034667 -0.241705206 Island 0.765158417 DDIT4 -963 chr14 24640947 24641852 -0.346838032 Island 0.704630646 REC8 0 chr17 5000803 5001047 -0.180094485 Island 0.639449923 ZFP3 -1867 chr13 88326244 88326244 -0.254835678 Island 0.622086898 SLITRK5 -1375 chr7 27188020 27188465 -0.198478999 Island 0.611495803 HOXA6 -652 chr1 46859671 46860511 -0.204828969 Island 0.603323126 FAAH 0 chr10 128593922 128594144 -0.21139227 Island 0.532062199 DOCK1 0 chr12 124247223 124247223 -0.291619587 Island 0.531144201 ATP6V0A2 -1636 chr22 38610376 38610795 -0.197239852 Island 0.519140873 MAFF -516 chr12 104697193 104697631 -0.159774024 Island 0.511530854 EID3 893 chrX 92928508 92928610 -0.215885822 Island 0.509900024 NAP1L3 0 chr17 56357994 56358318 0.235604062 Shore -2.771069849 MPO 0 chr2 47597118 47597331 0.170825315 Shore -1.989838284 EPCAM -652

66 chr12 112204756 112205368 -0.177217632 Shore 1.798343931 ALDH2 -6 chr12 112203801 112204506 -0.167361987 Shore 1.798343931 ALDH2 244 chr7 27205504 27205514 -0.189312475 Shore 1.750125912 HOXA9 -355 chr3 15372726 15372965 -0.191871292 Shore 1.691364717 SH3BP5 923 chr7 30028281 30028307 -0.439758955 Shore 1.682967495 SCRN1 1008 chr7 30027454 30027454 -0.299227828 Shore 1.682967495 SCRN1 1861 chr7 27184077 27184159 -0.156088057 Shore 1.652615374 HOXA5 -794 chr22 38201496 38201848 -0.328478194 Shore 1.54987965 H1F0 -242 chr8 30243930 30243930 -0.252405503 Shore 1.369854497 RBPMS -1888 chr7 100465051 100465833 -0.20478317 Shore 1.281381937 TRIP6 -78 chr7 27193351 27194013 -0.193087116 Shore 1.195082055 HOXA7 219 chr7 25018503 25018595 -0.16970931 Shore 0.966906711 OSBPL3 1066 chr7 27196759 27197239 -0.177214748 Shore 0.940816384 HOXA7 -463 chr1 220922046 220922217 -0.186478516 Shore 0.904301462 MOSC2 -436 chr1 67772896 67773044 -0.179679693 Shore 0.859823651 IL12RB2 2 chr3 49170496 49170794 0.161770178 Shore -0.78786347 LAMB2 0 chrY 21728575 21728575 0.23566283 Shore -0.772593829 CYorf15A 692 chr1 186650441 186650479 -0.254143241 Shore 0.757276969 PTGS2 -885 chr5 76249502 76250527 -0.21663642 Shore 0.713935248 CRHBP -634 chr6 31549563 31550090 -0.20860554 Shore 0.705812972 LTB 112 chr4 90757139 90757378 -0.206564689 Shore 0.683467526 SNCA -293

67 chr9 5831674 5831999 -0.28057155 Shore 0.672688294 ERMP1 -697 chrX 110039536 110039604 -0.240784813 Shore 0.667757542 CHRDL1 -543 chr4 156681475 156681475 -0.306895344 Shore 0.655209216 GUCY1B3 -1234 chr1 60280088 60280106 -0.186569105 Shore 0.648435017 HOOK1 488 chr18 47016218 47016218 -0.284177026 Shore 0.616623366 LOC729046 /// RPL17 -932 chr7 24614206 24614348 -0.286787283 Shore 0.557276298 MPP6 -1183 chr1 61549542 61549982 -0.239027386 Shore 0.555128781 NFIA -1563 chr12 93966060 93967711 -0.165376292 Shore 0.52397313 SOCS2 0 chr2 216877276 216877750 -0.197016908 Shore 0.50372197 MREG 565 chr21 37442759 37442777 -0.191372117 Shore 0.503048275 CBR1 -423 chr3 136539328 136539328 -0.246575275 Shore 0.501386314 TMEM22 -1351 chr10 45474317 45474372 -0.183899524 shelf 0.931044445 C10orf10 -60 chr1 156211409 156211434 -0.185683842 shelf 0.589711142 BGLAP 570 chr11 59823993 59824116 0.161461095 Open sea -2.706906127 MS4A3 14 chr6 133035379 133035379 -0.242403002 Open sea 1.849875377 VNN1 -191 chr6 31556255 31556255 -0.244242862 Open sea 1.1249658 LST1 -1279 chr13 48987165 48987165 0.238629161 Open sea -1.003641807 LPAR6 -562 chr14 106354912 106354912 -0.234479961 Open sea 0.985697199 FAM30A 1067 chr1 26644515 26645313 -0.164618365 Open sea 0.874440839 CD52 -31 chr4 159442782 159442782 -0.447897492 Open sea 0.781456346 RXFP1 117 chr10 98031125 98031337 -0.152269885 Open sea 0.696536347 BLNK 0

68 chr7 47611829 47611926 -0.220983979 Open sea 0.619420183 TNS3 -918 chrX 154563852 154563968 -0.175452155 Open sea 0.518822469 CLIC2 0 chr7 18535072 18535786 -0.206752311 Open sea 0.507854096 HDAC9 0

69 Table 2.7. Second DMR analysis to examine confounding effect of MLL cases

LSC epigenetic signature DMRs (p value<0.01) All Samples 84 3030 No MLL cases 49 1398 Overlap 73.5% (36/49) 77% (1076/1398)

70 Table 2.8. Ingenuity pathway analysis

Ingenuity Canonical Pathways -log(p-value) Molecules

Fatty Acid α-oxidation 2.8E00 ALDH2,PTGS2

Melatonin Degradation III 1.81E00 MPO

Anandamide Degradation 1.81E00 FAAH

NRF2-mediated Oxidative Stress Response 1.73E00 GSTM5,MAFF,CBR1

VDR/RXR Activation 1.59E00 BGLAP,HOXA10

Eicosanoid Signaling 1.57E00 PTGS2,PTGER4

Phenylethylamine Degradation I 1.47E00 ALDH2

HGF Signaling 1.35E00 DOCK1,PTGS2

Prostanoid Biosynthesis 1.34E00 PTGS2

Parkinson's Signaling 1.31E00 SNCA

Corticotropin Releasing Hormone Signaling 1.25E00 PTGS2,GUCY1B3

Granzyme A Signaling 1.22E00 H1F0

Glutathione Redox Reactions I 1.12E00 CLIC2

Relaxin Signaling 1.12E00 RXFP1,GUCY1B3

Aryl Hydrocarbon Receptor Signaling 1.12E00 GSTM5,NFIA

Histamine Degradation 1.09E00 ALDH2

eNOS Signaling 1.08E00 LPAR6,GUCY1B3

Tryptophan Degradation X (Mammalian, via Tryptamine) 1.08E00 ALDH2

71 Oxidative Ethanol Degradation III 1.08E00 ALDH2

Putrescine Degradation III 1.06E00 ALDH2

Ethanol Degradation IV 1.06E00 ALDH2

Triacylglycerol Degradation 1.03E00 FAAH

Phenylalanine Degradation IV (Mammalian, via Side Chain) 1.02E00 ALDH2

Gap Junction Signaling 1.01E00 BGLAP,GUCY1B3

IL-9 Signaling 9.95E-01 SOCS2

MIF-mediated Glucocorticoid Regulation 9.83E-01 PTGS2

Role of JAK2 in Hormone-like Cytokine Signaling 9.83E-01 SOCS2

Dopamine Degradation 9.5E-01 ALDH2

Production of Nitric Oxide and Reactive Oxygen Species in Macrophages 9.39E-01 MPO,HOXA10

Inhibition of Angiogenesis by TSP1 9.39E-01 GUCY1B3

Glutathione-mediated Detoxification 9.39E-01 GSTM5

IL-8 Signaling 9.35E-01 MPO,PTGS2

Endothelin-1 Signaling 9.35E-01 PTGS2,GUCY1B3

Sertoli Cell-Sertoli Cell Junction Signaling 9.35E-01 MPP6,GUCY1B3

ILK Signaling 9.32E-01 DOCK1,PTGS2

Ethanol Degradation II 8.99E-01 ALDH2

MIF Regulation of Innate Immunity 8.9E-01 PTGS2

FcγRIIB Signaling in B Lymphocytes 8.81E-01 BLNK

Primary Immunodeficiency Signaling 8.23E-01 BLNK

72 Noradrenaline and Adrenaline Degradation 8.23E-01 ALDH2

Lymphotoxin β Receptor Signaling 7.93E-01 LTB

Role of IL-17A in Arthritis 7.93E-01 PTGS2

Nur77 Signaling in T Lymphocytes 7.79E-01 HDAC9

Huntington's Disease Signaling 7.74E-01 HDAC9,SNCA

Colorectal Cancer Metastasis Signaling 7.6E-01 PTGS2,PTGER4

Phototransduction Pathway 7.59E-01 GUCY1B3

Role of JAK1 and JAK3 in γc Cytokine Signaling 7.46E-01 BLNK

Phospholipase C Signaling 7.43E-01 BLNK,HDAC9

Cell Cycle: G1/S Checkpoint Regulation 7.4E-01 HDAC9

CD40 Signaling 7.34E-01 PTGS2

Antiproliferative Role of Somatostatin Receptor 2 7.22E-01 GUCY1B3

Macropinocytosis Signaling 7.16E-01 RAB34

Role of MAPK Signaling in the Pathogenesis of Influenza 7E-01 PTGS2

IL-17 Signaling 6.94E-01 PTGS2

T Helper Cell Differentiation 6.94E-01 IL12RB2

JAK/Stat Signaling 6.94E-01 SOCS2

Glucocorticoid Receptor Signaling 6.92E-01 BGLAP,PTGS2

Growth Hormone Signaling 6.89E-01 SOCS2

Small Cell Lung Cancer Signaling 6.84E-01 PTGS2

STAT3 Pathway 6.84E-01 SOCS2

73 TREM1 Signaling 6.73E-01 MPO

Prolactin Signaling 6.73E-01 SOCS2

Cyclins and Cell Cycle Regulation 6.63E-01 HDAC9

Serotonin Degradation 6.63E-01 ALDH2

Superpathway of Melatonin Degradation 6.58E-01 MPO

ErbB Signaling 6.26E-01 EREG

Altered T Cell and B Cell Signaling in Rheumatoid Arthritis 6.17E-01 LTB

Crosstalk between Dendritic Cells and Natural Killer Cells 6.09E-01 LTB

FAK Signaling 6.09E-01 DOCK1

Neuregulin Signaling 5.97E-01 EREG

Chronic Myeloid Leukemia Signaling 5.97E-01 HDAC9

PPAR Signaling 5.93E-01 PTGS2

p53 Signaling 5.77E-01 HDAC9

Fcγ Receptor-mediated Phagocytosis in Macrophages and Monocytes 5.73E-01 DOCK1

Telomerase Signaling 5.73E-01 HDAC9

IGF-1 Signaling 5.73E-01 SOCS2

Paxillin Signaling 5.62E-01 DOCK1

Cholecystokinin/Gastrin-mediated Signaling 5.48E-01 PTGS2

Pancreatic Adenocarcinoma Signaling 5.41E-01 PTGS2

Type I Diabetes Mellitus Signaling 5.31E-01 SOCS2

Nitric Oxide Signaling in the Cardiovascular System 5.28E-01 GUCY1B3

74 Gαs Signaling 5.25E-01 PTGER4

Hereditary Breast Cancer Signaling 5.12E-01 HDAC9

Gα12/13 Signaling 5.09E-01 LPAR6

14-3-3-mediated Signaling 5.06E-01 SNCA

RhoA Signaling 4.91E-01 LPAR6

LXR/RXR Activation 4.8E-01 PTGS2

PI3K/AKT Signaling 4.8E-01 PTGS2

Ovarian Cancer Signaling 4.67E-01 PTGS2

PI3K Signaling in B Lymphocytes 4.67E-01 BLNK

Sperm Motility 4.61E-01 GUCY1B3

Type II Diabetes Mellitus Signaling 4.54E-01 SOCS2

Protein Kinase A Signaling 4.53E-01 PTGS2,H1F0

IL-12 Signaling and Production in Macrophages 4.51E-01 IL12RB2

Cellular Effects of Sildenafil (Viagra) 4.44E-01 GUCY1B3

Synaptic Long Term Depression 4.23E-01 GUCY1B3

CXCR4 Signaling 4.03E-01 DOCK1

Axonal Guidance Signaling 3.99E-01 DOCK1,ROBO3

Acute Phase Response Signaling 3.81E-01 SOCS2

Dopamine-DARPP32 Feedback in cAMP Signaling 3.73E-01 GUCY1B3

Role of NFAT in Regulation of the Immune Response 3.7E-01 BLNK

Dendritic Cell Maturation 3.66E-01 LTB

75 B Cell Receptor Signaling 3.62E-01 BLNK

Role of NFAT in Cardiac Hypertrophy 3.54E-01 HDAC9

Calcium Signaling 3.54E-01 HDAC9

Mitochondrial Dysfunction 3.5E-01 SNCA

Agranulocyte Adhesion and Diapedesis 3.47E-01 CD34

ERK/MAPK Signaling 3.45E-01 DOCK1

mTOR Signaling 3.4E-01 DDIT4

Hepatic Fibrosis / Hepatic Stellate Cell Activation 3.29E-01 LTB

Integrin Signaling 3.27E-01 DOCK1

Actin Cytoskeleton Signaling 3E-01 DOCK1

cAMP-mediated signaling 2.97E-01 PTGER4

LPS/IL-1 Mediated Inhibition of RXR Function 2.96E-01 GSTM5

Role of Osteoblasts, Osteoclasts and Chondrocytes in Rheumatoid Arthritis 2.94E-01 BGLAP

G-Protein Coupled Receptor Signaling 2.48E-01 PTGER4

Xenobiotic Metabolism Signaling 2.38E-01 GSTM5

Role of Macrophages, Fibroblasts and Endothelial Cells in Rheumatoid Arthritis 2.1E-01 LTB

76 Table 2.9. Ingenuity upstream regulator analysis

Upstream Regulator p-value of overlap Target molecules in dataset

ASXL1 8.18E-12 HOXA10,HOXA5,HOXA6,HOXA7,HOXA9

KAT6A 4.16E-07 HOXA5,HOXA7,HOXA9

NPM1 1.09E-06 HOXA10,HOXA5,HOXA7,HOXA9

phorbol myristate BGLAP,CD52,CRHBP,DDIT4,FAAH,HOXA5,HOXA7,HOXA9,IL12RB2,LTB,MPO,PTGER4,PTG acetate 1.88E-06 S2

KMT2A 6.98E-06 HOXA10,HOXA5,HOXA7,HOXA9

raloxifene 8.44E-06 BGLAP,BLNK,DDIT4,LAMB2,MPPED2,PTGS2

nimesulide 1.38E-05 EREG,MPO,PTGS2

1-methyl-4- phenylpyridinium 1.66E-05 DDIT4,MPO,SNCA

mir-223 2.19E-05 CD34,CRHBP,GSTM5,MREG,NFIA

EPZ004777 2.31E-05 HOXA10,HOXA9

BGLAP,BLNK,DDIT4,FAAH,GUCY1B3,HOXA10,HOXA9,LAMB2,LTB,MPO,MPPED2,OSBPL3, beta-estradiol 2.73E-05 PTGS2,RBPMS,SOCS2

diethylstilbestrol 2.93E-05 BGLAP,EREG,HOXA10,HOXA9,SOCS2,VNN1

EZH2 3.78E-05 HOXA10,HOXA6,HOXA7,HOXA9,LTB,PTGS2

PHF1 5.19E-05 HOXA10,HOXA6,HOXA9

RNF20 7.67E-05 HOXA10,HOXA9

MEN1 1.18E-04 BGLAP,HOXA7,HOXA9

77 paricalcitol 1.28E-04 BGLAP,PTGER4,PTGS2

bexarotene 1.53E-04 C10orf10,DDIT4,MAFF,PTGS2

PSIP1 1.60E-04 HOXA7,HOXA9

arsenite 2.08E-04 EREG,MAFF,MPO,PTGS2

HBP1 2.14E-04 H1F0,PTGS2

LIF 2.18E-04 BGLAP,CD34,EREG,PTGS2,SOCS2

IL12 (complex) 2.39E-04 FAAH,IL12RB2,LTB,PTGS2,SOCS2

staurosporine 2.58E-04 BGLAP,CPNE3,MPO,PTGS2

AHR 2.64E-04 ALDH2,CBR1,GSTM5,HDAC9,LTBP3,PTGS2

trans- hydroxytamoxifen 3.27E-04 BLNK,DDIT4,LAMB2,MPPED2

meloxicam 3.42E-04 MPO,PTGS2

3-methylcholanthrene 3.52E-04 BGLAP,GSTM5,PTGS2

beta-naphthoflavone 3.73E-04 CBR1,GSTM5,SOCS2

tretinoin 3.85E-04 BGLAP,CBR1,CD34,CD52,EPCAM,HOXA5,HOXA9,MPO,MPP6,MS4A3,PTGS2,REC8

miR-196a-5p (and other miRNAs w/seed AGGUAGU) 3.89E-04 FAM169A,HDAC9,HOOK1,HOXA5,HOXA7,HOXA9

RNF2 3.96E-04 HOXA5,HOXA7,REC8

rotenone 4.19E-04 MPO,PTGS2,SNCA

ALDH2,EREG,FAAH,GUCY1B3,HDAC9,IL12RB2,LST1,MAFF,MPO,NDFIP1,PTGER4,PTGS2,S lipopolysaccharide 4.48E-04 OCS2

5.00E-04 GUCY1B3,PTGS2 5'-

78 methylthioadenosine

PLA2G4A 5.00E-04 GSTM5,PTGS2

miR-221-3p (and other miRNAs w/seed GCUACAU) 5.37E-04 CLIC2,CPNE8,DDIT4,HOXA7,NDFIP1,OSBPL3,SLITRK5

RNA polymerase II 5.75E-04 BGLAP,BLNK,HOXA7,HOXA9,PTGS2

mir-196 5.89E-04 HOXA7,HOXA9

EREG 5.89E-04 EREG,PTGS2

infliximab 6.08E-04 DDIT4,IL12RB2,PTGS2

E. coli B5 lipopolysaccharide 6.79E-04 ALDH2,IL12RB2,MPO,PTGS2,SOCS2

NFATC2 6.79E-04 IL12RB2,PTGS2,SH3BP5,SOCS2

RELB 7.03E-04 LTB,MPO,PTGS2

doxycycline 7.37E-04 HOXA10,MPO,PTGS2

ESR1 7.85E-04 DDIT4,EREG,GUCY1B3,LTB,PTGS2,SOCS2

BDKRB2 7.91E-04 PTGS2,SNCA

dexmedetomidine 7.91E-04 DDIT4,PTGS2

CBFB 1.00E-03 MPO,PTGER4,SOCS2

TNF 1.02E-03 ALDH2,BGLAP,C10orf10,EREG,HDAC9,HOXA9,LTB,MAFF,MPO,PTGS2,RBPMS,SOCS2

miR-1243 (miRNAs w/seed ACUGGAU) 1.13E-03 BLNK,NAP1L3,SNCA

CSF3 1.14E-03 CD34,HOXA7,LTB,MPO

NPC2 1.15E-03 PTGER4,PTGS2

79 ceruletide 1.15E-03 MPO,PTGS2

forskolin 1.15E-03 BGLAP,CRHBP,DDIT4,EREG,GSTM5,HOXA5,PTGS2

dexamethasone 1.22E-03 BGLAP,C10orf10,CBR1,CRHBP,DDIT4,EREG,GSTM5,HOXA7,LAMB2,LTB,PTGS2,SOCS2

fumonisin B1 1.28E-03 LTB,PTGS2

genistein 1.30E-03 EREG,GUCY1B3,HOXA10,OSBPL3,PTGS2

EHF 1.37E-03 BLNK,EREG,VNN1

HOXA9 1.41E-03 CD34,CLIC2,HOXA9,NFIA

DOT1L 1.42E-03 HOXA10,HOXA9

CSF2 1.46E-03 LTB,MPO,PTGER4,PTGS2,REC8,SOCS2

stearic acid 1.56E-03 PTGS2,SNCA

Histone h3 1.62E-03 BGLAP,HOXA10,HOXA5,HOXA7,HOXA9

prostaglandin A1 1.72E-03 LPAR6,PTGS2

BMP15 1.72E-03 EREG,PTGS2

beta- glycerophosphoric acid 1.72E-03 BGLAP,PTGS2

CFTR 1.81E-03 BLNK,FAAH,PTGS2

SND1 1.88E-03 EREG,PTGS2

wortmannin 1.92E-03 BGLAP,PTGER4,PTGS2,SOCS2

COL18A1 2.00E-03 CD34,DDIT4,PTGS2

JAK 2.04E-03 PTGER4,SOCS2

TRIB3 2.04E-03 BGLAP,DDIT4

80 SP3 2.15E-03 BGLAP,EREG,IL12RB2,PTGS2

IFNG 2.20E-03 BLNK,C10orf10,FAAH,HDAC9,IL12RB2,LAMB2,LST1,LTB,PTGS2,SOCS2

CD28 2.25E-03 GUCY1B3,IL12RB2,PTGER4,PTGS2,SOCS2

cigarette smoke 2.27E-03 GSTM5,MAFF,PTGS2,SOCS2

FOLR1 2.34E-03 EPCAM,LAMB2,MPP6

AREG 2.40E-03 EREG,PTGS2

RUNX1 2.48E-03 BGLAP,CD34,MPO

Sos 2.49E-03 DOCK1,HOOK1,LTBP3,PTGS2

FOS 2.63E-03 BGLAP,DOCK1,EREG,HOOK1,LTBP3,PTGS2

dextran sulfate 2.76E-03 GSTM5,MPO,PTGS2,REC8

3-methyladenine 2.78E-03 PTGS2,SNCA grape seed extract 2.78E-03 MPO,PTGS2

butylated hydroxyanisol 2.78E-03 CBR1,GSTM5 thromboxane A2 2.80E-03 PTGS2

Gαq 2.80E-03 PTGS2

Jnkk 2.80E-03 PTGS2

olesoxime 2.80E-03 MPO

Pad2 2.80E-03 PTGS2

LPAR5 2.80E-03 PTGS2

PAP1 2.80E-03 PTGS2

81 FLT4 2.80E-03 BGLAP

Tpl2 kinase inhibitor 2.80E-03 PTGS2

Mucin 2.80E-03 PTGS2

PF-4523655 2.80E-03 DDIT4

cucurbitacin E 2.80E-03 PTGS2

dixanthogen 2.80E-03 GSTM5

EDN2 2.80E-03 PTGS2

CLN3 2.80E-03 HOOK1

SIGLEC7 2.80E-03 PTGS2

SIGLEC9 2.80E-03 PTGS2

WDR61 2.80E-03 HOXA9

ACSL4 2.80E-03 PTGS2

TFF1 2.80E-03 PTGS2

TLE6 2.80E-03 BGLAP

Pla2g2a 2.80E-03 PTGS2

LY311727 2.80E-03 PTGS2

BN 50730 2.80E-03 PTGS2

1-(1-glycero)dodeca- 1,3,5,7,9-pentaene 2.80E-03 PTGS2

CGP77675 2.80E-03 PTGS2

Ro 31-7549 2.80E-03 PTGS2

82 arachidic acid 2.80E-03 SNCA n-6 docosapentaenoic acid 2.80E-03 PTGS2

long-chain alcohol 2.80E-03 PTGS2

IL10 2.91E-03 C10orf10,FAAH,HDAC9,IL12RB2,PTGS2

HDAC4 2.95E-03 CHRDL1,HDAC9,PTGS2

PRKG1 2.98E-03 GUCY1B3,MPO

COMMD3-BMI1 2.98E-03 HOXA7,HOXA9

HOXA7 2.98E-03 CD34,HOXA7

NR3C2 3.03E-03 DDIT4,NFIA,PTGS2

TSC22D3 3.19E-03 BGLAP,PTGS2

miR-3976 (miRNAs w/seed AUAGAGA) 3.19E-03 CPNE3,HDAC9

miR-3186-5p (miRNAs w/seed AGGCGUC) 3.19E-03 CLIC2,FAM169A

UPF2 3.19E-03 MPPED2,PTGS2

taurine 3.19E-03 MPO,PTGS2 tetrachlorodibenzodio xin 3.21E-03 BLNK,CBR1,HDAC9,PTGS2,SOCS2

STAT6 3.22E-03 ALDH2,LTB,PTGS2,SOCS2

IFNA2 3.32E-03 IL12RB2,LPAR6,REC8,SOCS2

ERK 3.38E-03 BGLAP,EREG,MAFF,PTGS2

83 benzo(a)pyrene 3.38E-03 EREG,MAFF,MPO,PTGS2

Immunoglobulin 3.49E-03 LST1,MAFF,MPO,PTGS2

Histone h4 3.65E-03 BGLAP,HOXA9,PTGS2

melatonin 3.74E-03 BGLAP,MPO,PTGS2

FGF1 3.93E-03 GSTM5,LTB,PTGS2

miR-139-5p (miRNAs w/seed CUACAGU) 4.04E-03 DDIT4,HOXA9,NDFIP1,NFIA,SOCS2

PTGS1 4.08E-03 PTGER4,PTGS2

HBEGF 4.08E-03 EREG,PTGS2

diclofenac 4.08E-03 MPO,PTGS2

fulvestrant 4.14E-03 BGLAP,DDIT4,GUCY1B3,PTGS2

indomethacin 4.47E-03 CD34,MPO,PTGER4,PTGS2

Hsp27 4.57E-03 BGLAP,PTGS2

IFNE 4.57E-03 PTGER4,PTGS2

TGFB1 4.58E-03 ALDH2,BGLAP,CD34,DDIT4,EREG,LTBP3,MPP6,PTGER4,PTGS2,RBPMS,ROBO3

ATF4 4.65E-03 BGLAP,DDIT4,PTGS2

NFE2L2 4.82E-03 BGLAP,CBR1,GSTM5,MAFF,PTGS2

miR-1277-3p (and other miRNAs w/seed ACGUAGA) 4.82E-03 CPNE3,NDFIP1

IKZF1 4.87E-03 BLNK,DOCK1,SH3BP5

IKBKB 4.88E-03 EREG,LTB,PTGS2,SOCS2

84 ZBTB16 5.08E-03 BGLAP,CD34

AGTR1 5.08E-03 GSTM5,PTGS2

TXNIP 5.08E-03 DDIT4,PTGS2

H89 5.10E-03 BGLAP,EREG,PTGS2

IL27 5.21E-03 C10orf10,IL12RB2,PTGS2

miR-155-5p (miRNAs w/seed UAAUGCU) 5.31E-03 HOOK1,LPAR6,NDFIP1,NFIA,RAB34,ZFP3

STK11 5.33E-03 EREG,LTB,PTGS2

Growth hormone 5.45E-03 BGLAP,GSTM5,SOCS2

arbutin 5.59E-03 PTGS2

1-alpha,24(R),25- trihydroxyvitamin D3 5.59E-03 BGLAP

prostaglandin E3 5.59E-03 PTGS2

24R,25- dihydroxyvitamin D3 5.59E-03 BGLAP

des-Arg(10)-kallidin 5.59E-03 PTGS2

sPla2 5.59E-03 PTGS2

cyanidin 3-O- glucoside 5.59E-03 PTGS2

ASB2 5.59E-03 HOXA9

VRK2 5.59E-03 PTGS2

NOX5 5.59E-03 PTGS2

85 TMEM8B 5.59E-03 PTGS2

soy isoflavones 5.59E-03 PTGS2

3,4- dihydroxyphenyletha nol 5.59E-03 PTGS2

PTPRG 5.59E-03 CD34

PLA2G2F 5.59E-03 PTGS2

IDE 5.59E-03 SNCA

OSTF1 5.59E-03 BGLAP

ECSIT 5.59E-03 PTGS2

CBR1 5.59E-03 PTGS2

ARHGDIB 5.59E-03 PTGS2

SH3GLB2 5.59E-03 PTGS2

SC68376 5.59E-03 PTGS2

pyridoxal 5.59E-03 PTGS2

zileuton 5.59E-03 PTGS2

bifenthrin 5.59E-03 PTGS2

thiamin pyrophosphate 5.59E-03 MPO

benzylamine 5.59E-03 PTGS2

bumetanide 5.59E-03 PTGS2

incyclinide 5.59E-03 PTGS2

86 methylprednisolone acetate 5.59E-03 PTGS2

vanillic acid 5.59E-03 PTGS2

indolo(3,2- b)carbazole 5.59E-03 SOCS2

tetrahydropalmatine 5.59E-03 PTGS2

epoxyeicosatrienoic acid 5.59E-03 PTGS2

poly(ADP-ribose) 5.59E-03 PTGS2

laminaran 5.59E-03 PTGS2

lipooligosaccharide 5.59E-03 PTGS2

1beta,25- dihydroxyvitamin D3 5.59E-03 BGLAP

BAPTA-AM 5.61E-03 PTGS2,SOCS2

ICAM1 5.89E-03 IL12RB2,MPO

miR-3680-5p (miRNAs w/seed ACUCACU) 5.89E-03 HOOK1,RBPMS

miR-4716-5p (miRNAs w/seed CCAUGUU) 5.89E-03 NFIA,OSBPL3

nonylphenol 5.89E-03 EREG,PTGS2

IL18 6.32E-03 IL12RB2,MPO,PTGS2

ARHGAP21 6.46E-03 PTGS2,SH3BP5

N-acetylsphingosine 6.46E-03 GSTM5,PTGS2

87 AGT 6.68E-03 EREG,GSTM5,HOXA9,PTGS2,SNCA

luteolin 6.76E-03 MPO,PTGS2

WT1 6.99E-03 EREG,SH3BP5,SLC35G2

PPRC1 7.06E-03 DDIT4,PTGS2

APOA1 7.06E-03 MPO,PTGS2

VitaminD3-VDR- RXR 7.37E-03 BGLAP,HOXA10

EGFR 7.49E-03 EREG,HOXA7,PTGER4,PTGS2

baicalin 7.68E-03 BGLAP,CBR1

DUSP1 7.68E-03 IL12RB2,PTGS2

mifepristone 7.97E-03 DDIT4,EPCAM,HOXA10,PTGS2

EPO 8.07E-03 CD52,MPO,PTGS2,SOCS2 prostaglandin E2 8.27E-03 EREG,PTGER4,PTGS2,SOCS2

NOG 8.33E-03 BGLAP,PTGS2

DKK1 8.33E-03 BGLAP,EPCAM

sulprostone 8.37E-03 PTGS2

chloride 8.37E-03 PTGS2

W146 8.37E-03 PTGS2

imperatorin 8.37E-03 PTGS2

Eif2 8.37E-03 PTGS2

pycnogenols 8.37E-03 PTGS2

88 pinoresinol 8.37E-03 PTGS2

HN 8.37E-03 SH3BP5

RGD1560225 8.37E-03 PTGS2

SPEN 8.37E-03 BGLAP

CTSK 8.37E-03 PTGS2

ERC1 8.37E-03 PTGS2

AA-861 8.37E-03 PTGS2

amifostine 8.37E-03 PTGS2

trimetazidine 8.37E-03 MPO

tenidap 8.37E-03 PTGS2

acacetin 8.37E-03 PTGS2

flavone 8.37E-03 PTGS2

12- hydroxyeicosatetraen oic acid 8.37E-03 PTGS2

epiallopregnanolone 8.37E-03 PTGS2

IRF1 8.61E-03 IL12RB2,LTB,PTGS2

ZBTB20 8.66E-03 GSTM5,SOCS2

miR-517a-3p (and other miRNAs w/seed UCGUGCA) 8.66E-03 HOXA5,NFIA

miR-124-3p (and other miRNAs w/seed AAGGCAC) 8.98E-03 ATP6V0A2,CPNE3,ERMP1,HOXA5,NAP1L3,NDFIP1,NFIA,OSBPL3,RAB34,RHBDF1,SLITRK5

89 Gm-csf 9.00E-03 PTGS2,SOCS2

PDGFB 9.00E-03 BGLAP,PTGS2

FGF10 9.00E-03 PTGS2,VNN1

glutathione 9.00E-03 MPO,PTGS2 miR-1321 (and other miRNAs w/seed AGGGAGG) 9.20E-03 CD34,CHRDL1,HOXA10,OSBPL3

bucladesine 9.30E-03 BGLAP,FAAH,PTGER4,PTGS2

carbonyl cyanide m- chlorophenyl hydrazone 9.34E-03 DDIT4,PTGS2

2-deoxyglucose 9.69E-03 DDIT4,PTGS2

SRF 9.74E-03 GSTM5,MPPED2,MS4A3,PTGS2

zymosan 1.00E-02 C10orf10,PTGS2

BNIP3L 1.04E-02 BLNK,LTB

hemin 1.08E-02 GUCY1B3,PTGS2

miR-4793-5p (miRNAs w/seed CAUCCUG) 1.11E-02 ALDH2,NDFIP1

HDAC3 1.11E-02 CD34,PTGS2

Nfatc 1.11E-02 PTGS2

8-chloro-cAMP 1.11E-02 PTGS2

2-(3- hydroxypropoxy)calci triol 1.11E-02 BGLAP

90 NF-kappaB decoy 1.11E-02 PTGS2 trans-cinnamaldehyde 1.11E-02 PTGS2

tylophorine 1.11E-02 PTGS2

nebivolol 1.11E-02 PTGS2

RSPO3 1.11E-02 PTGS2

LEO1 1.11E-02 HOXA9

SGPL1 1.11E-02 PTGS2

ARID2 1.11E-02 BGLAP

LIN9 1.11E-02 BGLAP

TMEM119 1.11E-02 BGLAP

betulin 1.11E-02 PTGS2

PPP1R1B 1.11E-02 SNCA

ENPP1 1.11E-02 BGLAP

MUC2 1.11E-02 PTGS2

HSD11B2 1.11E-02 BGLAP

PHF19 1.11E-02 HOXA5

NUCB2 1.11E-02 PTGS2

TDO2 1.11E-02 PTGS2

TIA1 1.11E-02 PTGS2

UBE2D1 1.11E-02 PTGS2

AJUBA 1.11E-02 DOCK1

91 SKLB023 1.11E-02 PTGS2

farnesyl transferase 1.11E-02 MPO

SULT1E1 1.11E-02 PTGS2

ANXA6 1.11E-02 BGLAP

HDAC8 1.11E-02 HOXA5

furosemide 1.11E-02 PTGS2

1-butanol 1.11E-02 PTGS2

lestaurtinib 1.11E-02 MPO

lansoprazole 1.11E-02 PTGS2

7,8-dihydro-7,8- dihydroxybenzo(a)py rene 9,10-oxide 1.11E-02 PTGS2

ethyl linoleate 1.11E-02 PTGS2

cadmium sulfate 1.11E-02 PTGS2

polyphosphate 1.11E-02 BGLAP

vanadium pentoxide 1.11E-02 PTGS2

aclarubicin 1.11E-02 PTGS2

ecdysterone 1.11E-02 BGLAP

FOXL2 1.15E-02 MAFF,PTGS2

NFATC3 1.15E-02 DDIT4,PTGS2

miR-3944-5p (miRNAs w/seed GUGCAGC) 1.17E-02 CD34,NFIA,SH3BP5

92 miR-3974 (miRNAs w/seed AAGGUCA) 1.19E-02 ERMP1,MS4A3

seocalcitol 1.19E-02 BGLAP,PTGS2

Salmonella enterica serotype abortus equi lipopolysaccharide 1.21E-02 EREG,MAFF,PTGS2

PPARD 1.21E-02 ALDH2,PTGER4,PTGS2

8-bromo-cAMP 1.23E-02 ALDH2,BGLAP,PTGS2

PTHLH 1.23E-02 BGLAP,PTGS2

PDGF BB 1.24E-02 EREG,GUCY1B3,LAMB2,PTGS2

SP1 1.25E-02 BGLAP,EREG,IL12RB2,LTB,PTGS2

INHBA 1.27E-02 CPNE8,EREG,MPPED2

miR-3649 (miRNAs w/seed GGGACCU) 1.29E-02 CPNE3,HOXA10,MAFF

bortezomib 1.29E-02 BGLAP,PTGS2,SNCA

MAPK1 1.31E-02 BGLAP,OSBPL3,PTGS2,SOCS2

IRF2 1.31E-02 IL12RB2,PTGS2

miR-3115 (miRNAs w/seed UAUGGGU) 1.31E-02 HDAC9,PTGER4

miR-3928-3p (miRNAs w/seed GAGGAAC) 1.31E-02 GSTM5,OSBPL3

CD14 1.31E-02 MPO,PTGS2

INSR 1.33E-02 ALDH2,CBR1,EREG,SOCS2

93 STAT5A 1.35E-02 IL12RB2,LTB,SOCS2

15-deoxy-delta-12,14 -PGJ 2 1.37E-02 BGLAP,MPO,PTGS2

MAP3K14 1.39E-02 LTB,PTGS2

ILX-23-7553 1.39E-02 BGLAP

pyrophosphate 1.39E-02 BGLAP

nitrate 1.39E-02 MPO

dibutyryl cGMP 1.39E-02 PTGS2

dienogest 1.39E-02 PTGS2

Cpla2 1.39E-02 PTGS2

TET1 1.39E-02 HOXA9

elocalcitol 1.39E-02 BGLAP

GPR68 1.39E-02 PTGS2

carvacrol 1.39E-02 PTGS2

PARG 1.39E-02 PTGS2

Atf 1.39E-02 PTGS2

PLCE1 1.39E-02 PTGS2

CDH4 1.39E-02 PTGS2

PAPOLA 1.39E-02 PTGS2

DSPP 1.39E-02 BGLAP

FIGF 1.39E-02 BGLAP

94 PLAA 1.39E-02 PTGS2

2-cyclohexen-1-one 1.39E-02 PTGS2 pyridoxal phosphate- 6-azophenyl-2',4'- disulfonic acid 1.39E-02 PTGS2

mevalonolactone 1.39E-02 PTGS2

3-aminotriazole 1.39E-02 MPO

butylated hydroxytoluene 1.39E-02 PTGS2

NADPH 1.39E-02 PTGS2

5-hydroxytryptophan 1.39E-02 PTGS2

N-acetylglucosamine 1.39E-02 PTGS2

ginsenoside Rg1 1.39E-02 PTGS2

miR-144-3p (miRNAs w/seed ACAGUAU) 1.40E-02 CLIC2,HOXA10,HOXA7,NDFIP1,NFIA,PTGS2,SOCS2

progesterone 1.42E-02 FAAH,HOXA10,MPO,PTGER4,PTGS2

POR 1.43E-02 EREG,GSTM5,VNN1

IL17A 1.43E-02 EREG,MPO,PTGS2

miR-2355-3p (miRNAs w/seed UUGUCCU) 1.43E-02 ATP6V0A2,ERMP1

TBP 1.43E-02 HOXA9,PTGS2

ATP-gamma-S 1.52E-02 EREG,PTGS2

95 RARB 1.52E-02 HOXA5,PTGS2

CDX2 1.52E-02 HOXA5,HOXA9

SPIB 1.52E-02 BLNK,EPCAM

cobalt chloride 1.52E-02 DDIT4,PTGS2

L-triiodothyronine 1.52E-02 BGLAP,CBR1,PTGS2,RAB34

estradiol benzoate 1.56E-02 PTGER4,PTGS2

1-methyl-4-phenyl- 1,2,3,6- tetrahydropyridine 1.56E-02 MPO,PTGS2

IL4 1.58E-02 ALDH2,FAAH,IL12RB2,LTB,PTGS2,SOCS2

DOCK8 1.60E-02 PTGS2,SH3BP5

SPDEF 1.60E-02 CD34,LAMB2

MAP2K3 1.60E-02 PTGS2,RAB34

miR-448-3p (and other miRNAs w/seed UGCAUAU) 1.61E-02 DDIT4,MPPED2,NFIA,SLITRK5,SOCS2,ZFP3

miR-223-3p (miRNAs w/seed GUCAGUU) 1.64E-02 DDIT4,NFIA,SLC35G2,SNCA

EDN1 1.65E-02 EREG,PTGER4,PTGS2

miR-1839-5p (and other miRNAs w/seed AGGUAGA) 1.65E-02 HDAC9,HOXA7

nitrite 1.67E-02 MPO

teichoic acid 1.67E-02 PTGS2

96 LRRC26 1.67E-02 LTB

CACNA1C 1.67E-02 PTGS2

EN2 1.67E-02 SNCA

PDGFD 1.67E-02 PTGS2

PTGDR 1.67E-02 PTGS2

SSPN 1.67E-02 PTGS2

DES 1.67E-02 HOXA10

OGN 1.67E-02 BGLAP

TAB2 1.67E-02 PTGS2

PROK1 1.67E-02 PTGS2

PIM3 1.67E-02 IL12RB2

WAC 1.67E-02 BGLAP

glaucocalyxin A 1.67E-02 PTGS2

RNF17 1.67E-02 MPO

SNW1 1.67E-02 BGLAP

XPA 1.67E-02 PTGS2

PITPNA 1.67E-02 PTGS2

Pde4d 1.67E-02 PTGS2

ambroxol 1.67E-02 MPO

EGTA acetoxymethyl ester 1.67E-02 PTGS2

97 olmesartan medoxomil 1.67E-02 PTGS2

GW 5074 1.67E-02 PTGS2

4-nitroquinoline-1- oxide 1.67E-02 PTGS2 ferric nitrilotriacetate 1.67E-02 PTGS2

RWJ 67657 1.67E-02 PTGS2

dibenzoylmethane 1.67E-02 PTGS2

ibandronic acid 1.67E-02 BGLAP

dieldrin 1.67E-02 SNCA

propyl gallate 1.67E-02 PTGS2

phosphatidic acid 1.67E-02 PTGS2

phenylamil 1.67E-02 BGLAP canrenoate potassium 1.67E-02 PTGS2

miR-3615 (miRNAs w/seed CUCUCGG) 1.69E-02 CD34,SLITRK5

SASH1 1.69E-02 PTGS2,SH3BP5

ERG 1.69E-02 DOCK1,GUCY1B3,SNCA

SMAD4 1.76E-02 BGLAP,EREG,PTGS2

isotretinoin 1.83E-02 HOXA5,PTGS2

SP600125 1.86E-02 BGLAP,DDIT4,PTGS2

TRAF2 1.88E-02 EPCAM,RHBDF1

98 MET 1.92E-02 PTGS2,SH3BP5

GW9662 1.92E-02 MPO,PTGS2

ascorbic acid 1.92E-02 BGLAP,PTGS2

16,16- dimethylprostaglandi n E2 1.94E-02 PTGER4

(6)-gingerol 1.94E-02 PTGS2

Rhox4b (includes others) 1.94E-02 CD34

Glucocorticoid-GCR 1.94E-02 BGLAP

CTR9 1.94E-02 HOXA9

sphingomyelinase 1.94E-02 PTGS2

DVL1 1.94E-02 PTGS2

RAPGEF1 1.94E-02 PTGS2

ENO1 1.94E-02 PTGS2

TPSD1 1.94E-02 PTGS2

LPAR2 1.94E-02 PTGS2

APOBEC1 1.94E-02 PTGS2

SF3B2 1.94E-02 HOXA5

MLLT3 1.94E-02 HOXA9

IL18RAP 1.94E-02 PTGS2

PLA2G2A 1.94E-02 PTGS2

99 NEDD4 1.94E-02 SNCA

4-coumaric acid 1.94E-02 PTGS2

zimelidine 1.94E-02 C10orf10

eugenol 1.94E-02 PTGS2

atropine 1.94E-02 PTGS2

S-allyl-L-cysteine 1.94E-02 PTGS2

20alpha- hydroxycholesterol 1.94E-02 BGLAP

IL13 1.97E-02 CD52,PTGS2,SNCA,VNN1

TGFA 1.97E-02 EREG,PTGS2

miR-3180-3p (and other miRNAs w/seed GGGGCGG) 1.99E-02 DDIT4,NFIA,SLITRK5,ZFP3

GATA2 2.02E-02 CD34,MPO

miR-3691-5p (miRNAs w/seed GUGGAUG) 2.02E-02 CPNE8,CRHBP

FGFR1 2.02E-02 BGLAP,PTGS2

SNCA 2.02E-02 BGLAP,SNCA

imatinib 2.02E-02 BLNK,SOCS2

ESR2 2.03E-02 EREG,PTGS2,SOCS2

CEBPB 2.06E-02 BGLAP,BLNK,PTGER4,PTGS2

TRAF3 2.07E-02 EPCAM,RHBDF1

100 actinomycin D 2.08E-02 GSTM5,GUCY1B3,PTGS2

ciglitazone 2.12E-02 MPO,PTGS2

miR-4318 (miRNAs w/seed ACUGUGG) 2.17E-02 EPCAM,MREG

6-hydroxydopamine 2.17E-02 DDIT4,SNCA

eicosapentenoic acid 2.17E-02 PTGS2,SNCA

2',3'-dialdehyde ATP 2.22E-02 PTGS2

1L-6-hydroxymethyl- chiro-inositol 2-(R)- 2-O-methyl-3-O- octadecylcarbonate 2.22E-02 PTGS2

ganglioside GD1a 2.22E-02 PTGS2

ganglioside GM1 2.22E-02 PTGS2 desoxycorticosterone 2.22E-02 PTGS2

SMTNL1 2.22E-02 PTGS2

atrazine 2.22E-02 PTGS2 voltage-gated calcium channel 2.22E-02 CD34

Cyclin A 2.22E-02 CD34

BCG vaccine 2.22E-02 PTGS2

TCF/LEF 2.22E-02 PTGS2

RNF40 2.22E-02 BGLAP

EN1 2.22E-02 SNCA

101 TFRC 2.22E-02 SNCA

IL1R2 2.22E-02 MPO

GRIP1 2.22E-02 PTGS2

CBX8 2.22E-02 HOXA9

VEGFC 2.22E-02 PTGS2

PIM2 2.22E-02 IL12RB2

DLX5 2.22E-02 BGLAP

S1PR3 2.22E-02 PTGS2

PLA2G5 2.22E-02 PTGS2

WNT3 2.22E-02 PTGS2

naproxen 2.22E-02 PTGS2

allyl isothiocyanate 2.22E-02 PTGS2

NCX-4040 2.22E-02 PTGS2

capsazepine 2.22E-02 PTGS2

flufenamic acid 2.22E-02 PTGS2

lapatinib 2.22E-02 EREG

carnosol 2.22E-02 PTGS2

glimepiride 2.22E-02 BGLAP

acetic acid 2.22E-02 MPO

palmitoyl-Cys((RS)- 2,3-di(palmitoyloxy)- propyl)-Ala-Gly-OH 2.22E-02 PTGS2

102 rutin 2.22E-02 PTGS2

tosyllysine chloromethyl ketone 2.22E-02 PTGS2

mannan 2.22E-02 PTGS2

22(S)- hydroxycholesterol 2.22E-02 BGLAP

lead acetate 2.22E-02 GSTM5

MYB 2.22E-02 CD34,PTGS2

TBK1 2.22E-02 PTGS2,SH3BP5

MKL2 2.27E-02 GSTM5,MS4A3

miR-3622a-3p (and other miRNAs w/seed CACCUGA) 2.27E-02 ATP6V0A2,FAM169A

allopurinol 2.32E-02 MPO,PTGS2

miR-4438 (miRNAs w/seed ACAGGCU) 2.32E-02 HOXA5,SH3BP5

FBXO32 2.32E-02 HOOK1,PTGS2

VEGFA 2.33E-02 ALDH2,CD34,PTGS2

WNT3A 2.33E-02 BGLAP,DDIT4,PTGS2

okadaic acid 2.37E-02 BGLAP,PTGS2

TFAP2A 2.42E-02 EREG,PTGER4

Rxr 2.48E-02 BGLAP,PTGS2

NRAS 2.48E-02 CD34,PTGS2

103 daidzein 2.48E-02 OSBPL3,PTGS2

TCF 2.48E-02 EPCAM,PTGS2

ganglioside 2.49E-02 PTGS2

tannic acid 2.49E-02 PTGS2

ganglioside GT1 2.49E-02 PTGS2

RUNX1T1 2.49E-02 CD34

dovitinib 2.49E-02 LTB

FAT1 2.49E-02 PTGS2

NEK7 2.49E-02 PTGS2

SLC18A2 2.49E-02 SNCA

SOX8 2.49E-02 VNN1

mir-194 2.49E-02 SOCS2

CD209 2.49E-02 PTGS2

PRDX1 2.49E-02 PTGS2

NEK6 2.49E-02 PTGS2

TAF6 2.49E-02 HOXA9

BTC 2.49E-02 PTGS2

HPSE 2.49E-02 PTGS2

ammonium trichloro(dioxoethyle ne-O,O'-)tellurate 2.49E-02 MPO

bupivacaine 2.49E-02 MPO

104 domoic acid 2.49E-02 PTGS2

phenyl-N-tert- butylnitrone 2.49E-02 PTGS2

racemic flurbiprofen 2.49E-02 PTGS2

ketorolac 2.49E-02 PTGS2

mesalamine 2.49E-02 PTGS2

rhodioloside 2.49E-02 BGLAP

nickel chloride 2.49E-02 PTGS2 hexamethoxyflavone 2.49E-02 PTGS2

theophylline 2.49E-02 MPO

tyrphostin AG 127 2.49E-02 PTGS2

tyrphostin AG 1024 2.49E-02 PTGS2

saturated fatty acid 2.49E-02 PTGS2 arachidonyltrifluorom ethane 2.49E-02 PTGS2

L-N6-(1-iminoethyl)- lysine 2.49E-02 PTGS2

miR-342-5p (and other miRNAs w/seed GGGGUGC) 2.49E-02 CD34,LST1,NFIA,ZFP3

PPARG 2.49E-02 BGLAP,MPO,PTGS2,VNN1

PARP1 2.53E-02 PTGS2,SOCS2

fatty acid 2.53E-02 PTGS2,VNN1

105 MYD88 2.58E-02 MAFF,PTGS2,SH3BP5

TAL1 2.58E-02 DOCK1,MPO

phorbol esters 2.58E-02 HOXA5,PTGS2

camptothecin 2.59E-02 LAMB2,LST1,MAFF,PTGER4,PTGS2

PPARA 2.64E-02 ALDH2,PTGS2,SOCS2,VNN1

SAMSN1 2.69E-02 PTGS2,SH3BP5

aspirin 2.69E-02 MPO,PTGS2

EPHB1 2.76E-02 PTGS2

(+)-catechin 2.76E-02 PTGS2

Pdgfr 2.76E-02 EREG

FFAR4 2.76E-02 PTGS2

SGPP1 2.76E-02 PTGS2

NQO2 2.76E-02 PTGS2

FOXN1 2.76E-02 MREG

mir-130 2.76E-02 HOXA5

NMNAT1 2.76E-02 SOCS2

RYBP 2.76E-02 HOXA7

PTAFR 2.76E-02 PTGS2

VIM 2.76E-02 BGLAP

DEFB103A/DEFB10 3B 2.76E-02 PTGS2

106 TAS1R3 2.76E-02 DDIT4

edaravone 2.76E-02 PTGS2

methylene blue 2.76E-02 MPO

clomipramine 2.76E-02 C10orf10

cinnamaldehyde 2.76E-02 PTGS2

rosmarinic acid 2.76E-02 PTGS2

morin 2.76E-02 PTGS2

midostaurin 2.76E-02 PTGS2

8-oxo-7- hydrodeoxyguanosine 2.76E-02 PTGS2

allopregnanolone 2.76E-02 PTGS2

formononetin 2.76E-02 BGLAP

miR-502-5p (and other miRNAs w/seed UCCUUGC) 2.80E-02 MARC2,MREG salmonella minnesota R595 lipopolysaccharides 2.86E-02 MAFF,PTGS2 bisindolylmaleimide I 2.91E-02 BGLAP,PTGS2

tamoxifen 2.95E-02 C10orf10,MAFF,PTGS2

ERBB2 2.95E-02 CD34,EREG,LTBP3,PTGS2,RAB34

miR-224-5p (miRNAs w/seed AAGUCAC) 2.96E-02 CPNE8,HOXA5,MAFF,NFIA

107 miR-128-3p (and other miRNAs w/seed CACAGUG) 2.98E-02 HOXA10,HOXA5,HOXA9,LPAR6,MPPED2,NFIA,RAB34

TRIM24 3.03E-02 BLNK,SOCS2

L-histidine 3.04E-02 DDIT4

estriol 3.04E-02 HOXA10

DNA- methyltransferase 3.04E-02 PTGS2

CARD9 3.04E-02 MPO

Agtr1b 3.04E-02 PTGS2

SYK/ZAP 3.04E-02 PTGS2

PP2A 3.04E-02 PTGS2

Rac 3.04E-02 PTGS2

trypsin 3.04E-02 PTGS2

Adaptor protein 1 3.04E-02 PTGS2

T 0070907 3.04E-02 PTGS2

Collagen Alpha1 3.04E-02 PTGS2

IL13RA2 3.04E-02 VNN1

PLD1 3.04E-02 PTGS2

MTA3 3.04E-02 BLNK

MAPK10 3.04E-02 PTGS2

ITGA3 3.04E-02 PTGS2

108 MZF1 3.04E-02 CD34

HNRNPAB 3.04E-02 PTGS2

sesamin 3.04E-02 PTGS2

manumycin A 3.04E-02 PTGS2

amiloride 3.04E-02 PTGS2

enterolactone 3.04E-02 BGLAP

phospholipid 3.04E-02 PTGS2

puerarin 3.04E-02 PTGS2

tyrphostin AG 1296 3.04E-02 PTGS2

miR-200b-3p (and other miRNAs w/seed AAUACUG) 3.05E-02 ATP6V0A2,CHRDL1,CRHBP,DDIT4,HOOK1,HOXA5,NFIA

fenofibrate 3.07E-02 HOXA7,MPO,PTGS2

Ca2+ 3.07E-02 BGLAP,FAAH,PTGS2

GNA15 3.08E-02 GUCY1B3,LAMB2

GW501516 3.14E-02 PTGER4,VNN1

hyaluronic acid 3.20E-02 MPO,PTGS2

CD3 3.20E-02 GUCY1B3,IL12RB2,PTGER4,PTGS2,SOCS2

hydrogen peroxide 3.22E-02 BGLAP,DDIT4,PTGER4,PTGS2

NFKBIA 3.22E-02 EREG,GSTM5,HOXA10,PTGS2

MAPT 3.23E-02 ERMP1,GSTM5,SNCA

F2 3.23E-02 EREG,HDAC9,PTGS2

109 cephaloridine 3.26E-02 DDIT4,GSTM5

bisphenol A 3.26E-02 EREG,PTGS2

miR-3189-5p (miRNAs w/seed GCCCCAU) 3.27E-02 FAM169A,MPPED2,SLITRK5

PD173074 3.31E-02 LTB

NFkB (family) 3.31E-02 PTGS2

perhexiline 3.31E-02 C10orf10

chlorcyclizine 3.31E-02 C10orf10

HRH1 3.31E-02 PTGS2

STAT1/3/5 dimer 3.31E-02 SOCS2

harmine 3.31E-02 PTGS2

astragalin 3.31E-02 PTGS2

SLAMF1 3.31E-02 IL12RB2

PLA2G10 3.31E-02 PTGS2

MIR101 3.31E-02 PTGS2

mir-214 3.31E-02 PTGS2 miR-643 (miRNAs w/seed CUUGUAU) 3.31E-02 EREG

KLF10 3.31E-02 BGLAP

TRPC1 3.31E-02 SNCA

NPPC 3.31E-02 GUCY1B3

110 10-nitrooleate 3.31E-02 PTGS2

GNA13 3.31E-02 PTGS2

PDCD4 3.31E-02 PTGS2

RHOB 3.31E-02 PTGS2

ebselen 3.31E-02 PTGS2

N-(3- oxododecanoyl)- homoserine lactone 3.31E-02 PTGS2

dimethyl fumarate 3.31E-02 PTGS2

D-sphingosine 3.31E-02 PTGS2

SKF-38393 3.31E-02 BGLAP

N,N- dimethylsphingosine 3.31E-02 PTGS2

buthionine sulfoximine 3.31E-02 PTGS2

progestin 3.31E-02 HOXA10

NR1I3 3.32E-02 GSTM5,MAFF

KAT5 3.32E-02 EREG,HOXA9

NR3C1 3.38E-02 BGLAP,DDIT4,LTB,PTGS2,SNCA

INHA 3.38E-02 EREG,PTGS2

cyclic AMP 3.40E-02 BGLAP,EPCAM,PTGS2

miR-4687-3p (miRNAs w/seed GGCUGUU) 3.44E-02 C10orf10,HDAC9

111 JAK2 3.44E-02 MPO,PTGS2

RB1 3.50E-02 BGLAP,H1F0,ROBO3

miR-361-5p (miRNAs w/seed UAUCAGA) 3.54E-02 ATP6V0A2,GUCY1B3,NFIA

MKL1 3.56E-02 GSTM5,MS4A3

RARA 3.56E-02 HOXA5,PTGS2

leukotriene C4 3.58E-02 PTGS2

SIM2 3.58E-02 ROBO3

TRPV1 3.58E-02 PTGS2

LPAR1 3.58E-02 PTGS2

CSF3R 3.58E-02 MPO

CD46 3.58E-02 PTGER4

PTGFR 3.58E-02 PTGS2

PTGER1 3.58E-02 PTGS2

PLD2 3.58E-02 PTGS2

SP7 3.58E-02 BGLAP

FABP1 3.58E-02 FAAH

Go6983 3.58E-02 PTGS2

1,2- dimethylhydrazine 3.58E-02 PTGS2

mevastatin 3.58E-02 PTGS2

112 SU5402 3.58E-02 LTB

icatibant 3.58E-02 PTGS2

farnesyl pyrophosphate 3.58E-02 PTGS2 gamma-linolenic acid 3.58E-02 PTGS2

prostaglandin 3.58E-02 PTGS2

auranofin 3.58E-02 PTGS2

IL5 3.68E-02 LTB,RBPMS,SOCS2

CCND1 3.75E-02 CPNE3,EREG,MAFF

miR-381-3p (and other miRNAs w/seed AUACAAG) 3.76E-02 ATP6V0A2,HOXA9,NDFIP1,NFIA,OSBPL3,SNCA miR-3198 (and other miRNAs w/seed UGGAGUC) 3.81E-02 IL12RB2,SCRN1 miR-4421 (and other miRNAs w/seed CCUGUCU) 3.81E-02 H1F0,MANSC1

HDAC2 3.81E-02 CD34,PTGS2

A23187 3.81E-02 DDIT4,PTGS2

sulforafan 3.81E-02 GSTM5,PTGS2

prostaglandin A2 3.85E-02 LPAR6

lonafarnib 3.85E-02 PTGS2

cyclooxygenase 3.85E-02 PTGS2

113 IKK (complex) 3.85E-02 PTGS2

dihydroartemisinin 3.85E-02 PTGS2

KLF9 3.85E-02 HOXA10

mir-135 3.85E-02 HOXA10

PDGFRA 3.85E-02 BGLAP

PRDX2 3.85E-02 PTGS2

BAG1 3.85E-02 PTGS2

salmeterol 3.85E-02 DDIT4

hydrogen sulfide 3.85E-02 PTGS2

N1,N11- diethylnorspermine 3.85E-02 PTGS2

myricetin 3.85E-02 PTGS2

neuroprotectin D1 3.85E-02 PTGS2

gambogic acid 3.85E-02 PTGS2

Cd2+ 3.85E-02 PTGS2

manganese 3.85E-02 PTGS2

paraquat 3.93E-02 PTGS2,SNCA

miR-26a-5p (and other miRNAs w/seed UCAAGUA) 3.96E-02 ERMP1,HOOK1,HOXA5,HOXA9,MREG,PTGS2

corticosterone 4.06E-02 DDIT4,PTGS2

IL21 4.06E-02 IL12RB2,SOCS2

114 miR-3103-5p (and other miRNAs w/seed GAGGGAG) 4.11E-02 ALDH2,CD34,CHRDL1

epoprostenol 4.12E-02 PTGS2

cyclic GMP 4.12E-02 IL12RB2

TMSB4 4.12E-02 BGLAP

1'-acetoxychavicol acetate 4.12E-02 PTGS2

IGFBP7 4.12E-02 PTGS2

MAPK8IP1 4.12E-02 SNCA

PLCG2 4.12E-02 PTGS2

CD83 4.12E-02 PTGS2

S1PR2 4.12E-02 PTGS2

SP2 4.12E-02 BGLAP

RPS6KA5 4.12E-02 PTGS2

ITGAL 4.12E-02 IL12RB2

RND3 4.12E-02 PTGS2

CD47 4.12E-02 IL12RB2

pyridoxine 4.12E-02 PTGS2

rofecoxib 4.12E-02 PTGS2

caffeic acid 4.12E-02 PTGS2

lauric acid 4.12E-02 PTGS2

115 resveratrol 4.27E-02 BGLAP,MPO,PTGS2 miR-4300 (and other miRNAs w/seed GGGAGCU) 4.30E-02 GSTM5,HOXA7,MPPED2

POU2F1 4.32E-02 DDIT4,GSTM5

lovastatin 4.32E-02 MPP6,PTGS2

misoprostol 4.38E-02 PTGS2

RNASEL 4.38E-02 PTGS2

SALL4 4.38E-02 EPCAM

CGB (includes others) 4.38E-02 PTGS2

PRKCI 4.38E-02 PTGS2

HDC 4.38E-02 PTGS2

SCGB1A1 4.38E-02 PTGS2

superoxide 4.38E-02 PTGS2

allyl sulfide 4.38E-02 GSTM5

D609 4.38E-02 PTGS2

amlodipine 4.38E-02 PTGS2

piperine 4.38E-02 PTGS2

2- arachidonoylglycerol 4.38E-02 PTGS2

leptomycin B 4.38E-02 PTGS2

acadesine 4.38E-02 PTGS2

116 Alpha catenin 4.39E-02 EREG,PTGS2

BMP7 4.39E-02 BGLAP,CD34

miR-340-5p (miRNAs w/seed UAUAAAG) 4.41E-02 CPNE8,GUCY1B3,H1F0,HOXA10,MAFF,MPPED2,NFIA,RBPMS

IL1RN 4.52E-02 HDAC9,PTGS2

miR-324-5p (miRNAs w/seed GCAUCCC) 4.58E-02 HOXA9,MPPED2,ZFP3

interferon beta-1a 4.59E-02 LTB,SNCA

5-azacytidine 4.59E-02 MAFF,PTGS2

miR-130a-3p (and other miRNAs w/seed AGUGCAA) 4.62E-02 EREG,HOXA5,MPPED2,NAP1L3,NFIA,RAB34

leukotriene B4 4.65E-02 MPO

calcipotriene 4.65E-02 BGLAP

Cyclin E 4.65E-02 CD34

NMDA Receptor 4.65E-02 PTGS2

HCAR2 4.65E-02 PTGS2

WWTR1 4.65E-02 BGLAP

TEAD1 4.65E-02 PTGS2

Endothelin 4.65E-02 PTGS2

fontolizumab 4.65E-02 IL12RB2

p85 (pik3r) 4.65E-02 PTGS2

117 CBX4 4.65E-02 HOXA7

FKBP4 4.65E-02 HOXA10

miR-4766-5p (and other miRNAs w/seed CUGAAAG) 4.65E-02 ATP6V0A2

WNT7A 4.65E-02 HOXA10

TACR1 4.65E-02 PTGS2

S1PR1 4.65E-02 PTGS2

PD 169316 4.65E-02 PTGS2

ellagic acid 4.65E-02 PTGS2

ethoxyquin 4.65E-02 GSTM5

NO 1886 4.65E-02 PTGS2

ethyl pyruvate 4.65E-02 PTGS2

dipyridamole 4.65E-02 PTGS2

isoliquiritigenin 4.65E-02 PTGS2

magnolol 4.65E-02 PTGS2

ammonium chloride 4.65E-02 SNCA

vanadate 4.65E-02 BGLAP

miR-335-5p (and other miRNAs w/seed CAAGAGC) 4.65E-02 NDFIP1,OSBPL3,SCRN1

SMAD7 4.73E-02 HDAC9,LTBP3

HGF 4.75E-02 EPCAM,GUCY1B3,PTGS2,SOCS2

118 miR-133a-3p (and other miRNAs w/seed UUGGUCC) 4.75E-02 CPNE3,HOXA9,NFIA,RAB34,SNCA

Pdgf Ab 4.92E-02 EREG

3M-002 4.92E-02 PTGS2

artesunic acid 4.92E-02 PTGS2

miR-4652-3p (miRNAs w/seed UUCUGUU) 4.92E-02 CPNE8

HOXB9 4.92E-02 EREG

CREB3L1 4.92E-02 BGLAP benzyl isothiocyanate 4.92E-02 PTGS2

S-(2,3- bispalmitoyloxypropy l)-cysteine- GDPKHPKSF 4.92E-02 PTGS2

IL2 4.93E-02 IL12RB2,LTB,PTGS2,SOCS2

TLR4 4.94E-02 EREG,PTGS2,SH3BP5 miR-7a-5p (and other miRNAs w/seed GGAAGAC) 4.96E-02 DDIT4,NFIA,OSBPL3,SNCA

119 Table 2.10. Association of LSC epigenetic signature with DMRs of genetic mutations

Columns are chromosome, start, end, diffMethyl(difference of methylation percentage for LSC-Blast), diffexp( difference of expression log2 value for LSC-Blast), Gene, DMR name, DNMT3A,IDH1,IDH2,TET2,NPM1,ASXL1(Genetic mutations tested here to look at overlap of LSC epigenetic signature and DMR for mutations), Mechanism(which mechanism regulate each DMR).

"1" represent overlap and "0" represents no overlap between the corresponding LSC epigenetic signature and DMRs for

DNMT3A, IDH1, IDH2, TET2, NPM1 and ASXL1.

Gene DMR name DNMT3A IDH1 IDH2 TET1 TET2 NPM1 ASXL1 Mechanism

MPO MPO/DMR1 0 0 0 0 0 0 1 UpstremRegulator

HOXA9 HOXA9/DMR1 0 0 0 0 0 0 0 PrimaryEpi

HOXA9 HOXA9/DMR2 0 0 0 0 0 0 0 PrimaryEpi

HOXA9 HOXA9/DMR3 0 0 0 0 0 1 0 UpstremRegulator

HOXA9 HOXA9/DMR4 0 0 0 0 0 1 0 UpstremRegulator

HOXA9 HOXA9/DMR5 0 0 0 0 0 0 0 PrimaryEpi

SCRN1 SCRN1/DMR1 1 0 0 0 0 0 0 EpigeneticEnzyme

CD34 CD34 0 0 0 0 0 0 1 UpstremRegulator

120 CPNE3 CPNE3 0 0 0 0 0 0 0 PrimaryEpi

HOXA10 HOXA10 0 0 0 0 0 1 0 UpstremRegulator

MANSC1 MANSC1 0 0 0 0 0 0 0 PrimaryEpi

NDFIP1 NDFIP1 0 0 0 0 0 0 0 PrimaryEpi

FAM24B FAM24B 0 0 0 1 0 0 0 EpigeneticEnzyme

FAM169A FAM169A 1 0 0 0 0 0 0 EpigeneticEnzyme

MPPED2 MPPED2 1 0 0 0 0 0 0 EpigeneticEnzyme

PTGER4 PTGER4 0 0 0 0 0 0 0 PrimaryEpi

HOXA7 HOXA7/DMR1 0 0 0 0 0 0 0 PrimaryEpi

HOXA7 HOXA7/DMR2 1 0 0 0 0 1 0 CommonTarget

RHBDF1 RHBDF1 1 0 0 0 0 0 0 EpigeneticEnzyme

EREG EREG 0 0 0 0 0 0 0 PrimaryEpi

CPNE8 CPNE8 0 0 0 0 0 1 0 UpstremRegulator

RAB34 RAB34/DMR1 0 0 0 0 0 0 0 PrimaryEpi

LTBP3 LTBP3 0 0 0 0 0 0 0 PrimaryEpi

RAB34 RAB34/DMR2 0 0 0 0 0 0 0 PrimaryEpi

ROBO3 ROBO3 0 0 0 0 0 0 1 UpstremRegulator

GSTM5 GSTM5 0 0 0 0 0 0 1 UpstremRegulator

DDIT4 DDIT4 0 0 0 0 0 0 1 UpstremRegulator

REC8 REC8 1 0 0 0 1 0 1 CommonTarget

ZFP3 ZFP3 0 0 0 0 0 0 0 PrimaryEpi

121 SLITRK5 SLITRK5 0 0 0 0 0 0 0 PrimaryEpi

HOXA6 HOXA6 1 0 0 0 0 1 0 CommonTarget

FAAH FAAH 0 0 0 0 0 0 1 UpstremRegulator

DOCK1 DOCK1 0 1 0 0 1 1 0 CommonTarget

ATP6V0A2 ATP6V0A2 1 0 0 0 0 0 0 EpigeneticEnzyme

MAFF MAFF 0 0 0 0 0 0 0 PrimaryEpi

EID3 EID3 1 0 0 0 0 0 0 EpigeneticEnzyme

NAP1L3 NAP1L3 0 0 0 0 0 0 0 PrimaryEpi

MPO MPO/DMR2 0 0 0 0 0 0 1 UpstremRegulator

EPCAM EPCAM 0 0 1 1 0 0 0 EpigeneticEnzyme

ALDH2 ALDH2/DMR1 0 0 0 0 0 0 0 PrimaryEpi

ALDH2 ALDH2/DMR2 0 0 0 0 0 0 0 PrimaryEpi

HOXA9 HOXA9/DMR6 0 0 0 0 0 0 0 PrimaryEpi

SH3BP5 SH3BP5 1 0 0 0 0 0 0 EpigeneticEnzyme

SCRN1 SCRN1/DMR2 0 0 0 0 1 0 0 EpigeneticEnzyme

SCRN1 SCRN1/DMR3 1 0 0 0 0 0 0 EpigeneticEnzyme

HOXA5 HOXA5 0 0 0 0 0 1 0 UpstremRegulator

H1F0 H1F0 0 0 0 0 0 0 1 UpstremRegulator

RBPMS RBPMS 0 0 0 0 0 0 0 PrimaryEpi

TRIP6 TRIP6 0 0 0 0 0 0 0 PrimaryEpi

HOXA7 HOXA7/DMR3 1 0 0 1 0 1 0 CommonTarget

122 OSBPL3 OSBPL3 0 0 0 0 0 0 0 PrimaryEpi

HOXA7 HOXA7/DMR4 0 0 0 0 0 1 0 UpstremRegulator

MOSC2 MOSC2 0 0 0 0 0 0 0 PrimaryEpi

IL12RB2 IL12RB2 0 0 0 0 0 0 0 PrimaryEpi

LAMB2 LAMB2 0 1 0 0 0 0 0 EpigeneticEnzyme

CYorf15A CYorf15A 0 0 0 0 0 0 0 PrimaryEpi

PTGS2 PTGS2 0 0 0 0 0 0 0 PrimaryEpi

CRHBP CRHBP 0 1 0 0 0 0 1 CommonTarget

LTB LTB 1 0 0 0 0 0 0 EpigeneticEnzyme

SNCA SNCA 0 0 0 0 0 0 0 PrimaryEpi

ERMP1 ERMP1 1 0 0 0 0 0 0 EpigeneticEnzyme

CHRDL1 CHRDL1 0 0 0 0 0 0 0 PrimaryEpi

GUCY1B3 GUCY1B3 0 0 0 0 0 0 0 PrimaryEpi

HOOK1 HOOK1 0 0 0 0 0 0 0 PrimaryEpi

LOC729046 /// RPL17 RPL17 0 0 0 0 0 0 0 PrimaryEpi

MPP6 MPP6 1 0 0 0 0 0 0 EpigeneticEnzyme

NFIA NFIA 0 0 0 0 0 0 0 PrimaryEpi

SOCS2 SOCS2 0 0 0 0 0 0 1 UpstremRegulator

MREG MREG 0 0 0 0 0 0 0 PrimaryEpi

CBR1 CBR1 0 0 0 0 0 0 0 PrimaryEpi

TMEM22 TMEM22 0 0 0 0 0 0 0 PrimaryEpi

123 C10orf10 C10orf10 0 0 0 0 0 0 0 PrimaryEpi

BGLAP BGLAP 0 0 0 0 0 0 0 PrimaryEpi

MS4A3 MS4A3 0 1 1 1 0 0 1 CommonTarget

VNN1 VNN1 1 0 0 0 0 1 0 CommonTarget

LST1 LST1 1 0 0 0 0 1 0 CommonTarget

LPAR6 LPAR6 0 0 0 0 0 0 0 PrimaryEpi

FAM30A FAM30A 0 0 0 0 0 0 1 UpstremRegulator

CD52 CD52 1 0 0 0 0 0 0 EpigeneticEnzyme

RXFP1 RXFP1 0 0 0 0 0 0 0 PrimaryEpi

BLNK BLNK 0 0 0 0 0 1 0 UpstremRegulator

TNS3 TNS3 1 0 0 0 0 1 0 CommonTarget

CLIC2 CLIC2 0 0 0 0 1 1 0 CommonTarget

HDAC9 HDAC9 0 0 0 0 0 0 0 PrimaryEpi

19 4 2 4 4 15 13

124 Table 2.11. Multivariate analysis of overall survival of TCGA patients using either DNA methylation or gene expression

DNA Methylation Gene Expression

Variable HR(95% CI) p HR(95% CI) p

Group 1.9(1.2-2.9) 0.003 1.7(1.0-2.7) 0.03

Age 1.04(1.03-1.06) 8.5X 10-7 1.04(1.02-1.06) 1 X 10-6

Cytogenetic risk Intermediate vs low 2.7(1.3-5.2) 0.005 2.2(1.0-4.5) 0.04

High vs low 2.7(1.3-5.6) 0.006 2.2(1.0-4.7) 0.05

NPM1 0.8(0.5-1.3) 0.39 1.0(0.6-1.7) 1.00

FLT3 1.7(1.1-2.8) 0.03 1.5(0.9-2.5) 0.10

125 Table 2.12. Univariate overall survival analysis for LSC epigenetic signature regarding differential gene expression in various cohorts

TCGA Metzeler et al Wouters et al Wilson et al Variable HR (95% CI) p HR (95% CI) p HR (95% CI) p HR (95% CI) p LSC score 2.4 (1.6-3.6) 1x10-5 1.9 (1.3-2.8) 1 x 10-3 2.3 (1.7-3.1) 2x10-7 2.2 (1.6-3.1) 2x10-6 (High vs. Low)* 1.03 (1.01- 1.03 (1.02- Age 1.04 (1.03-1.06) 1x10-9 3 x 10-4 1.01 (1.0-1.03) 3 x 10-2 4x10-6 1.04) 1.05) Cytogenetics Intermediate 2.7 (1.4-5.2) 2 x 10-3 - - 2.8 (1.7-4.7) 3x10-5 1.9 (0.9-4.1) 1.1 x 10-1 vs. low High vs. low 3.9 (1.9-7.8) 2 x 10-4 - - 4.7 (2.7-8.4) 1x10-7 4.2 (1.9-9.6) 6x 10-4 FLT3 1.1 (0.7-1.7) 7.1 x 10-1 2.2 (1.5-3.3) 8x10-5 1.8 (1.3-2.5) 5 x 10-4 1.1 (0.8-1.6) 5.2 x 10-1 NPM1 1.4 (0.9-2.1) 1.8 x 10-1 0.8 (0.5-1.2) 2.4 x 10-1 0.9 (0.6-1.3) 5.1 x 10-1 0.8 (0.5-1.1) 2.0 x 10-1 *LSC score is determined as described in method

126 Table 2.13. Multivariate overall survival analysis for LSC epigenetic signature regarding differential gene expression in various cohorts

Metzeler et al Wouters et al Wilson et al

Variable HR (95% CI) p HR (95% CI) p HR (95% CI) p LSC score 1.6 (1.0-2.4) 4 x 10-2 1.8 (1.2-2.6) 3 x 10-3 1.8 (1.2-2.6) 5 x 10-3 (High vs. Low)* 1.03 (1.01- 1.02 (1.01- Age 4 x 10-4 1.02 (1.0-1.03) 3 x 10-2 2 x 10-3 1.04) 1.04) Cytogenetics Intermediate vs. low - - 2.4 (1.4-4.2) 2 x 10-3 1.4 (0.6-3.3) 3.9 x 10-1 High vs. low - - 3.1 (1.6-5.9) 7 x 10-4 3.0 (1.2-7.1) 2 x 10-2 FLT3 2.3 (1.5-3.6) 1 x 10-4 1.8 (1.3-2.6) 2 x 10-3 1.6 (1.0-2.5) 4 x 10-2 NPM1 0.7 (0.5-1.1) 8 x 10-2 0.5 (0.3-0.7) 2 x 10-4 0.7 (0.5-1.1) 1.2 x 10-1 *LSC score is determined as described in method

127 Table 2.14. Univariate overall survival analysis for genetic mutations in epigenome modifying enzymes in TCGA

Genetic mutation HR (95% CI) p DNMT3A 1.8 (1.2-2.7) 0.004 IDH1 0.8 (0.4-1.5) 0.4 IDH2 1.0 (0.6-1.9) 0.9 TET2 0.8 (0.4-1.7) 0.6 ASXL1 2.0 (0.6-6.4) 0.2

128 Table 2.15. Multivariate overall survival analysis including DNMT3A mutation for LSC epigenetic signature in TCGA

DNA Methylation Gene Expression Variable HR(95% CI) p HR(95% CI) p Group 1.9 (1.2-3.0) 0.005 1.7 (1.0-2.7) 0.04 Age 1.0 (1.0-1.0) 9.3 x 10-7 1.0 (1.0-1.0) 1.2x10-6 Cytogenetic risk Intermediate/Normal 2.7 (1.3-5.6) 0.007 2.2 (1.0-4.6) 0.04 High 2.8 (1.3-5.9) 0.007 2.2 (1.0-4.8) 0.06 NPM1 0.8 (0.5-1.4) 0.46 1.0 (0.6-1.7) 0.99 FLT3 1.7 (1.0-2.8) 0.04 1.5 (0.9-2.5) 0.1 DNMT3 1.0 (0.6-1.6) 1.0 1.0 (0.7-1.6) 0.92

129 Table 2.16. Multivariate overall survival analysis for LSC epigenetic signature within intermediate cytogenetic risk patients in TCGA

DNA Methylation Gene Expression

Variable HR (95% CI) p HR (95% CI) p

Group 1.8 (1.0-3.1) 0.05 1.9 (1.1-3.3) 0.03

Age 1.0 (1.0-1.1) 3 x 10-4 1.0 (1.0-1.1) 2 x 10-4

NPM1 0.7 (0.4-1.3) 0.2 0.9 (0.9-1.1) 0.7

FLT3 2.5 (1.3-4.8) 4 x 10-3 2.3 (1.2-4.2) 0.01

DNMT3A 1.0 (0.6-1.8) 0.9 1.1 (0.7-1.9) 0.7

130 Table 2.17. Antibodies for Flow Cytometry

Cell Catalog Working Surface Fluorophore Manufacturer Application Number Dilution marker CD3 APC-Cy7 341090 CD19 555414 1:50 PE-Cy5 CD20 BD 555624 To sort LSPCs CD34 APC Bioscience 340667 1:50 from AML CD38 PE-Cy7 335790 1:100 CD90 PE 555596 1:25 CD3 APC-Cy7 341090 1:50 CD19 APC 555415 1:50 BD Bioscience CD33 PE 555450 1:50 To test chimerism/ CD45 PB 560367 1:50 engraftment of LSC frequency in CD45.1 PE-Cy7 25-0453-82 1:100 NSG mice (mouse) eBioscience Ter119 PE-Cy5 15-5921-83 1:100 (mouse)

131 Table 2.18. Primers used for sequencing of TET2, IDH1, IDH2, and DNMT3A mutations of AML

Primers Sequence 5' to 3' Size Tm Reference

(1) TET2 exon 3 PCR1 F TGAACTTCCCACATTAGCTGGT 955 55 Gelsi-Boyer et al. British Journal of Haematology (2) TET2 exon 3 PCR1 R GAAACTGTAGCACCATTAGGCATT 2009, 145(6): 788-800 (3) TET2 exon 3 PCR1 Seq GATAGAAATAAACACATTTT (4) TET2 exon 3 PCR2 F CAAAAGGCTAATGGAGAAAGACGTA 836 55 (5) TET2 exon 3 PCR2 R GCAGAAAAGGAATCCTTAGTGAACA (6) TET2 exon 3 PCR3 F GCCAGTAAACTAGCTGCAATGCTAA 846 55 (7) TET2 exon 3 PCR3 R TGCCTCATTACGTTTTAGATGGG (8) TET2 exon 3 PCR4 F GACCAATGTCAGAACACCTCAA 867 60 (9) TET2 exon 3 PCR4 R TTGATTTTGAATACTGATTTTCACCA (10) TET2 exon 3 PCR5 F TTGCAACATAAGCCTCATAAACAG 788 60 (11) TET2 exon 3 PCR5 R ATTGGCCTGTGCATCTGACTAT (12) TET2 exon 3 PCR6 F GCAACTTGCTCAGCAAAGGTACT 781 60 (13) TET2 exon 3 PCR6 R TGCTGCCAGACTCAAGATTTAAAA (14) TET2 exon 4 F ATACTACATATAATACATTCTAATTCCCTCACTG 495 55 (15) TET2 exon 4 R TGTTTACTGCTTTGTGTGTGAAGG (16) TET2 exon 5 F CATTTCTCAGGATGTGGTCATAGAAT 286 55 (17) TET2 exon 5 R CCCAATTCTCAGGGTCAGATTTA (18) TET2 exon 6 F AGACTTATGTATCTTTCATCTAGCTCTGG 599 60 (19) TET2 exon 6 R ACTCTCTTCCTTTCAACCAAAGATT (20) TET2 exon 7 F ATGCCACAGCTTAATACAGAGTTAGAT 362 55 (21) TET2 exon 7 R TGTCATATTGTTCACTTCATCTAAGCTAAT

132 (22) TET2 exon 8 F GATGCTTTATTTAGTAATAAAGGCACCA 354 55 (23) TET2 exon 8 R TTCAACAATTAAGAGGAAAAGTTAGAATAATATTT (24) TET2 exon 9 F TGTCATTCCATTTTGTTTCTGGATA 361 55 (25) TET2 exon 9 R AAATTACCCAGTCTTGCATATGTCTT (26) TET2 exon 10 F CTGGATCAACTAGGCCACCAAC 774 55 (27) TET2 exon 10 R CCAAAATTAACAATGTTCATTTTACAATAAGAG (28) TET2 exon 11 PCR1 F GCTCTTATCTTTGCTTAATGGGTGT 748 60 (29) TET2 exon 11 PCR1 R TGTACATTTGGTCTAATGGTACAACTG (30) TET2 exon 11 PCR2 F AATGGAAACCTATCAGTGGACAAC 1107 60 (31) TET2 exon 11 PCR2 R TATATATCTGTTGTAAGGCCCTGTGA (32) IDH1 exon 4 F TGTGTTGAGATGGACGCCTATTTG 481 55 Thol F et al. Haematologica 2010, (33) IDH1 exon 4 R TGCCACCAACGACCAAGTCA 95(10): 1668-1674 (34) IDH2 exon 4 F GGGGTTCAAATTCTGGTTGA 290 53 (35) IDH2 exon 4 R CTAGGCGAGGAGCTCCAGT (36) DNMT3A exons 7-8 F ATGGTCCCCTTGAGTGTCAG 836 56 Fernandez-Mercado et al. PLoS One 2012, 7(8): (37) DNMT3A exons 7-8 R CATCACCCCAATTCCAGACT e42334 (38) DNMT3A exons 9-10 F CTGTATCTGGTCCCCTCCAG 747 56 (39) DNMT3A exons 9-10 R CTCCCTAAGCATGGCTTTCC (40) DNMT3A exons 11-12 F GGGAACAAGTTGGAGACCAG 490 56 (41) DNMT3A exons 11-12 R GGTCCCATGTCATTCAAACC (42) DNMT3A exon 13 F GTCACAGTGCCTCCCTTTTC 308 56 (43) DNMT3A exon 13 R TGGACACAGTCAGCCAGAAG (44) DNMT3A exon 14 F CAGGGCTTAGGCTCTGTGAG 359 56 (45) DNMT3A exon 14 R AGGTGTGCTACCTGGAATGG (46) DNMT3A exons 15-16 F CGGTCTTTCCATTCCAGGTA 614 56

133 (47) DNMT3A exons 15-16 R CATCATTTCGTTTTGCCAGA (48) DNMT3A exon 17 F GACTTGGGCCTACAGCTGAC 345 58 (49) DNMT3A exon 17 R CAAAATGAAAGGAGGCAAGG (50) DNMT3A exons 18-19 F CTTCCTGTCTGCCTCTGTCC 552 56 (51) DNMT3A exons 18-19 R ATGAAGCAGCAGTCCAAGGT (52) DNMT3A exons 19b-20 GCAGCACTGTGCAATATGGT 549 56 F (53) DNMT3A exons 19b-20 CTTCCCCACTATGGGTCATC R (54) DNMT3A exons 21 F GCGGGGAGTTTGAAGAGAGT 342 56 (55) DNMT3A exons 21 R CCACACTAGCTGGAGAAGCA (56) DNMT3A exons 22 F TTTGGTAGACGCATGACCAG 301 56 (57) DNMT3A exons 22 R CAGGACGTTTGTGGAAAACA (58) DNMT3A exons 23 F TCCTGCTGTGTGGTTAGACG 654 56 (59) DNMT3A exons 23 R CCTCTCTCCCACCTTTCCTC (60) DNMT3A exon 17 F CCTCGATGTCCTTACTATGGATACTCCA 402 63 Additional primers designed to cover the ones (61) DNMT3A exon 17 R CAAGGGCTGCCTCCAGGTGCTGAG 69 not working in previous (62) DNMT3A exon 17 F CTCACCTGCCGAGACCAG 276 59 rows, all the three new pairs worked on DNMT3A (63) DNMT3A exon 17 R CCTCCAGGTGCTGAGTGTG 60 exon 17 (48) DNMT3A exon 17 F GACTTGGGCCTACAGCTGAC 437 60 (64) DNMT3A exon 17 R TTTGCCCTTTACCCTCTCAA 57

Note: For IDH1 and IDH2, a single point mutation was tested in exon 4 (R132 and R140 respectively); for TET2 and DNMT3A mutations, multiple exons were tested based on regions of frequent somatic mutation according to COSMIC database (Wellcome Trust Sanger Institute).

134 Chapter 3

Epigenetic basis of human normal hematopoietic development

135 This work is an ongoing project of the Feinberg Lab and Johns Hopkins. All publication rights are reserved for these institutions and the presentation of this work here does not preclude future publication elsewhere.

Summary

DNA methylation plays an indispensable role during tissue development and cellular differentiation. Hematopoiesis is one of best-understood and characterized developmental processes. We hypothesized that DNA methylation would be an essential mechanism during lineage differentiation in human hematopoiesis. Here, we provide a comprehensive methylome map of HSPCs, particularly in myeloid lineage, by performing a genome-wide DNA methylation analysis. We found that DNA methylation distinguished distinct lineages of hematopoiesis, suggesting a critical role of it during hematopoietic development. In concordance with previous studies for tissue development, we observed that most of DNA methylation changes during hematopoietic differentiation occurred at non-CpG island regions such as CpG island shores or open seas. We identified DMRs in potential novel regulators of hematopoiesis including HMHB1 and

MIR539, as well as in previously known genes such as MPO and CDK6. The DMRs for normal hematopoiesis were enriched in a regulatory genomic element, super-enhancer.

We found massive epigenetic variation between murine and human hematopoiesis, explaining the intrinsic differences during the hematopoietic development in those species.

136 Results

Comprehensive DNA Methylation Analysis Shows Tight Clustering of Human

Hematopoietic Stem and Progenitor Cells

Human hematopoiesis proceeds through a series of multipotent and oligopotent stem and progenitor cells that progressively lose self-renewal ability and become more restricted in their differentiation potential (Figure 3.1). These critical functional properties are mediated in part through epigenetic mechanisms, including DNA methylation. We obtained bone marrow from five normal donors and isolated HSPC by fluorescence-activated cell sorting (FACS) including: hematopoietic stem cells (HSC), multipotent progenitors (MPP), L-MPP, common myeloid progenitors (CMP), megakaryocyte/erythroid progenitors (MEP), and GMP (Figure 3.2, Tables 3.1 and 3.2).

In order to further understand epigenetic variation during early human hematopoiesis, we generated genome-scale methylation profiles for normal hematopoietic stem and progenitor cell populations. Strikingly, multidimensional scaling analysis utilizing the top 1000 most variable CpG positions revealed tight clustering of human HSPC populations by lineage with no outliers (Figure 3.3). As the distance between clusters in multidimensional scaling is a measure of their similarity, this analysis indicates that DNA methylation reflects the function and hierarchy of the HSPC populations. For example, HSC and MPP clusters are close together reflecting their functional similarities. Similarly, L-MPP are located between HSC/MPP and GMP clusters, but farther away from MEP, supporting the hypothesis that L-MPP is an early lymphoid progenitor, which retains myeloid programs for GMP, but not MEP, differentiation (Doulatov et al., 2010; Goardon et al., 2011).

137 DMR Analysis Identified Previously Known and Novel Regulators in Human

Hematopoietic Development

The DMRs identified across HSPC not only permit clustering of these populations, but also have the potential to reveal novel regulators of hematopoietic lineage development. We first examined the DMRs for genes already known to play such a role.

The DMR analysis identified myeloperoxidase (MPO), a well-established protein involved in neutrophil activity (Klebanoff, 2005), which showed progressive hypomethylation going from HSC to GMP, but exhibited hypermethylation in MEP

(Figure 3.4). Similarly, CDK6, a cyclin-dependent protein kinase important in hematopoietic cell differentiation (Kozar and Sicinski, 2005; Malumbres et al., 2004), was progressively hypermethylated during differentiation from HSC to MEP and GMP

(Figure 3.4). In both cases, these results were confirmed through direct pyrosequencing of these loci (Figure 3.4), validating the DNA methylation array approach. In addition to these two examples, a number of genes known to be involved in early hematopoiesis were found to be differentially methylated including: STAT3, KEL, IDH2, HLF,

NOTCH1, GATA1, and HOX family genes (Table 3.3, see Appendix2).

This analysis also identified novel sites of epigenetic variation during hematopoiesis. For example, HMHB1, encoding one of the minor histocompatibility antigens, was found to be hypomethylated in L-MPP and GMP, suggesting a possible role in GMP differentiation (Figure 3.5). Progressive hypomethylation was also identified in MIR539 going from HSC to MEP, suggesting that this microRNA may contribute to erythropoiesis (Figure 3.5). Interestingly, the MIR539 gene is located in DLK1-DIO3 imprinting region that contains a miRNA cluster involved in leukemia pathogenesis

138 (Benetatos et al., 2013). Further validation of these novel candidate regulators will require functional experiments.

We next sought to classify the DMRs among normal HSPCs according to their global genomic location in islands, shores, shelves, and open seas. Focusing on the comparison of HSC with GMP, we found that most DMRs were not in CpG islands, but were enriched outside of the islands, predominantly in shelves and open seas, compared to the distribution of random length matched DMRs (Figure 3.6a). The comparison of

HSC with MEP also showed enrichment of DMRs at non-CpG island regions such as shores, shelves, and open seas (Figure 3.6b). Similarly, genes with an inverse correlation between DNA methylation and gene expression were located outside of the islands themselves, with the strongest correlation at shores and open seas for both comparisons

(Figure 3.7). In addition to these comparisons, more than 50% of the DMRs among

HSPCs were in open seas (Table 3.4). Thus, functional epigenetic differences during early human hematopoietic differentiation occur in CpG sparse regions, consistent with other recent studies of differentiation (Irizarry et al., 2009; Ji et al., 2010) and cancer

(Hansen et al., 2011; Irizarry et al., 2009).

DMRs for Normal Hematopoiesis are Enriched in Super-enhancers

In order to further investigate a direct mechanistic link between DNA methylation change and regulation of hematopoietic development, we examined the overlap of DMRs that we identified for normal hematopoiesis in DNA regulatory elements, super- enhancers. Super-enhancers are clusters of transcriptional enhancers bound by mater transcription factors, and associated with genes that define cell identity and tissue types

(Hnisz et al., 2013; Whyte et al., 2013). From a published study that has identified super-

139 enhancer in 86 different human cell and tissue samples, we selected three categories of

tissue types; first, normal cell types related to hematopoietic development such as CD34+

(hematopoietic progenitor cells) and CD14+ (monocytes), second, cell lines related to

hematological disorders including jurkat, K562 and MM1S, third, other normal tissues or

cell lines including H1, adipose, angular gyrus, and spleen. First, we examined the

enrichment of DMRs from normal hematopoiesis including HSC vs GMP and HSC vs

MEP comparisons, in super-enhancers of the different tissues or cell types. Remarkably,

DMRs for GMP differentiation were significantly enriched in relevant cell types

including monocytes and CD34+ cell population (Table 3.5). We identified master transcription factors showing differential methylation as HSC differentiates into GMP, such as ZNF217, known to be involved in cell proliferation (Figure 3.8a and Table 3.6, see Appendix3). Among the DMRs of HSC vs GMP comparison, contained within super- enhancers of monocytes, 27.9% was located in monocyte (CD14+ cells) specific super- enhancers (Table 3.6, see Appendix3). For example, a DMR of TREM1 that encodes a

receptor expressed on monocytes was located in a monocyte specific super-enhancer and

showed hypomethylation along with upregulation in GMP (Figure 3.8b and Table 3.6,

see Appendix3). In addition to the relevant cell types, the DMRs of HSC vs GMP were

overlapped with super-enhancers of jurkat and MM1S cells whereas, DMRs

distinguishing MEP from HSC are enriched in K562 cells, implying involvement of

epigenetic deregulation of the regulatory region in disease pathogenesis.

Human Hematopoiesis Displays Distinct Epigenetic Regulation Compared to Murine

Hematopoiesis

140 In addition to these genomic regional changes, there was an overall global change in the level of DNA methylation in human hematopoiesis (Figure 3.9 and Table 3.4).

Global hypomethylation was observed upon MEP differentiation from CMP, but not in

GMP differentiation, and most DMRs distinguishing MPP from CMP were less methylated in CMP in human hematopoiesis (Figure 3.9 and Table 3.4). We also found that all DMRs lost methylation when L-MPP differentiates into GMP (Figure 3.9 and

Table 3.4). Previous epigenetic study of mouse hematopoiesis (Ji et al., 2010) allowed us to compare human and mouse hematopoietic differentiation for two pairs, MPP vs CMP and CMP vs GMP. We examined overlap of genes of DMR lists of the two comparisons.

In both comparisons, most of the genes were different with only 4-5% of total genes in human overlapping with mouse: 7 out of 166 genes of MPP vs CMP, and 30 out of 728 genes of CMP vs GMP (Table 3.7, see Appendix4). Among the 7 and 30 common genes for the two comparisons, 5 and 14 of them showed the opposite direction of methylation changes for the MPP vs CMP and CMP vs GMP pairs, respectively. For example,

VKORC1L1 that encodes a subunit of the vitamin K epoxide reductase complex, was hypermethylated when CMP differentiate into GMP in human, but hypomethylated in the

CMP to GMP transition in mouse (Table 3.7, see Appendix4). These results suggest that there is great variability in the specific epigenetic sites regulating mouse and human hematopoiesis.

Discussion

The role of DNA methylation in hematopoietic development has been implicated in several studies that have shown an indispensable role of DNMT1, DNMT3A, and

141 DNMT3B in hematopoiesis (Broske et al., 2009; Challen et al., 2012; Tadokoro et al.,

2007; Trowbridge et al., 2009). In addition, dynamic DNA methylation changes during

hematopoietic development in mouse and human have been demonstrated (Bartholdy et

al., 2014; Bock et al., 2012; Ji et al., 2010). In order to provide a comprehensive

methylome map of human hematopoiesis, particularly myeloid lineage, we analyzed

genome-wide DNA methylation of a complete set of HSPCs in myeloid lineage. Our

study suggested several significant findings for the epigenetic basis of human

hematopoietic development. First, global hypomethylation is a core mechanism of hematopoietic differentiation, as MPP to CMP, CMP to MEP, and L-MPP to GMP differentiations were accompanied by loss of methylation (Figure 3.9). In addition to identifying previously known and novel candidate genes during hematopoietic differentiation, we found that DNA methylation were likely to occur at regulatory elements such as super-enhancer, suggesting a critical role of epigenetics in gene expression of lineage specific genes. Furthermore, we found distinct epigenetic plasticity between human and mouse hematopoiesis. We observed few overlaps of genes with

DNA methylation changes between mouse and human hematopoiesis. Besides the

distinct study design and experimental platform between the studies, intrinsically distinct

mechanisms exist in hematopoietic development in mouse and human. For example,

human blood contains 3 to 5-fold more neutrophils than mouse, while mouse blood has 2

fold more lymphocytes, indicating differences in the generation of specific types of

hematopoietic progenitor cells (Mestas and Hughes, 2004). Furthermore, mouse

hematopoietic system is composed of a different set of HSPCs, and a different set of cell surface markers for HSPCs are used to isolate each population. For example, multiple

142 subpopulations constitute an MPP compartment, and Flk2 and Slamf1 are used to differentiate those subpopulations (Bock et al., 2012; Chao et al., 2008; Doulatov et al.,

2012; Ji et al., 2010).

We could compare our results to previous DNA methylation study of human hematopoiesis, and found few common observations, yet there were many distinct and novel findings in this study. We observed global hypomethylation when MPPs differentiated into CMPs, as Bartholdy et al. has reported. However, we identified global hypomethylation in CMP to MEP transition, while Bartholdy et al. has observed balanced hyper and hypomethylation in this differentiation (Bartholdy et al., 2014). In addition, we noticed that DMR lists from all pair-wise comparisons among HSPCs (p <0.01) of our study significantly differed from 561 loci distinguishing LT-HSC, ST-HSC, CMP, and

MEP, identified in Bartholdy et al. 51 out of 3613 (1.4%) DMRs of our lists including

DMRs in KLF1 and GATA1, overlapped with the 561 loci from Bartholdy et al. The discordance between our study and Bartholdy et al. may come from two major sources: a different set of HSPCs and methods used in those studies. Our study includes two more progenitor populations, L-MPP and GMP, beside HSC, MPP (called as ST-HSC in

Bartholdy et al.), CMP, and MEP. We have used 450K array which interrogate ~48000

CpGs across the genome without a bias toward CpG islands and gene promoters, while

HELP_Promoter array, used in Bartholdy et al., examines ~26000 loci, primarily targeting gene promoters.

In summary, we provide a comprehensive methylome map of a complete set of

HSPCs in myeloid lineage, demonstrating the role of DNA methylation at regulatory

143 elements during lineage differentiation and dynamic epigenetic plasticity distinguishing

human hematopoiesis from murine hematopoietic development.

Materials and Methods

Human Samples

Fresh human bone marrow mononuclear cells (BMMC) from healthy donors (2 x 108

cells per donor, Catalog#: ABM006) were purchased from ALLCELLS® (Emeryville,

CA). A CD34+ cell-enrichment step was performed with the human progenitor cells enrichment kit with CD61 depletion (Stem Cell Technologies, Canada, Catalog # 19356) on a RobSep machine from the same company. PBMC or BMMC were separated with

Ficoll-Paque Plus (Amersham Biosciences, Piscataway, NJ, Catalog number: 17-1440-

03), and cryopreserved in 1 x freezing medium (90%FBS + 10%DMSO).

Flow Cytometry Analysis and Cell Sorting

A battery of antibodies (Abs) was used for staining, analysis and sorting of progenitor cells from either healthy BMMCs, as well as lineage analysis human chimerism/engraftment (Table 3.2). Cells were either analyzed or sorted using a FACS

Aria II cytometer (BD Biosciences, Franklin Lakes, NJ). Analysis of flow cytometry raw data was done with FlowJo Software (Treestar, Ashland, OR).

Illumina Infinium Human Methylation 450 Bead Array Assay

Genomic DNA from each sample was purified using the MasterPure DNA purification kit (Epicentre) according to the manufacturer’s protocol. The genomic DNA (250-500ng) was treated with sodium bisulfate using the Zymo EZ DNA Methylation Kit (ZYMO

Research) as recommended by the manufacturer, with the alternative incubation

144 conditions for the Illumina Infinium Methylation Assay. Converted DNA was eluted in

11ul of elution buffer. DNA methylation level was measured using Illumina Infinium HD

Methylation Assay (Illumina) according to the manufacturer’s specifications.

Methylation array data are deposited at the Gene Expression Omnibus (GEO) with

accession number GSE63409.

Illumina Infinium Human Methylation 450 Bead Array Analysis

Raw intensity files were obtained using minfi package (Aryee et al., 2014) to calculate methylation ratios (Beta values). The data was normalized using Illumina preprocessing method implemented in minfi. Several quality control measures were applied to remove

arrays with low quality. Control probes were examined on the 450k array to assess several measures including bisulfite conversion, extension, hybridization, specificity and others. One of the MPP samples (BM2712) showed low quality for the measures, so they

were removed for further analysis. Next, median methylated and unmethylated signals

were calculated for each arrays; no array was identified for signal values lower than 10.5.

For multidimensional scaling analysis, probes containing an annotated SNP (dbSNP137)

at the single-base extension or CpG sites were removed (17398 probes removed). Minfi

1.8.9 was used.

Bump hunting method previously described was applied to identify DMRs in 450k array

(Aryee et al., 2014; Jaffe et al., 2012). Beta value of 0.1 (10% of methylation difference)

was used as cutoff when finding DMRs. Statistical significance was assigned by

permutations testing and the P-value cutoff used for downstream analysis was <0.01 that

corresponded to Benjamini-Hochberg adjusted p-value <0.1 (data not shown) unless

different cutoff was designated in result part. Bumphunter 1.2.0 was used.

145 Bisulfite Pyrosequencing

100ng of genomic DNA from each sample was treated with sodium bisulfate using an EZ

DNA methylation Gold Kit (ZYMO research) following manufacturer’s protocol. The

bisulfate treated DNA was PCR amplified using unbiased nested primers. Quantitative

pyrosequencing was performed using a PSQ HS96 (Biotage) to validate DMR regions.

The DNA methylation percentage at each CpG site was measured using the Q-CpG

methylation software (Biotage). SssI treated human genomic DNA was used as 100%

methylated controls and human genomic DNA amplified by Repli-G mini kit (Qiagen)

was used as the non-methylated (0%) DNA control. Table 3.8 provides the primer sequence used for the pyrosequencing reactions with the chromosomal coordinates in the

University of California at Santa Cruz February 2009 human genome assembly (hg19) for each CpG site investigated.

Affymetrix Microarray Expression Analysis

Total RNA was extracted from each FACS-sorted cell population using RNeasy® Plus

Mini (QIAGEN, Valencia, CA, Catalog#: 74134) according to the manufacture’s

protocol. All RNA samples were quantified with 2100 Bioanalyzer (Agilent

Technologies, Santa Clara, CA), subjected to reverse transcription, two consecutive

rounds of linear amplification, and production and fragmentation of biotinylated cRNA.

15µg of cRNA from each sample was hybridized to HG U133 Plus 2.0 microarrays.

Hybridization and scanning were performed according to the manufacture’s instruction

(Affymetrix). This step was performed at the PAN center of Stanford University. Data

were normalized by GC robust multi-array average method and analyzed on

R/Bioconductor. BM2770 GMP, BM 2759 L-MPP, BM2761 CMP, BM2770 CMP were

146 removed from further analysis due to low quality (GEO GSE63270). University of

California at Santa Cruz February 2009 human genome assembly (hg19) for each CpG site investigated.

Affymetrix Microarray Expression Analysis

Total RNA was extracted from each FACS-sorted cell population using RNeasy® Plus

Mini (QIAGEN, Valencia, CA, Catalog#: 74134) according to the manufacture’s protocol. All RNA samples were quantified with 2100 Bioanalyzer (Agilent

Technologies, Santa Clara, CA), subjected to reverse transcription, two consecutive rounds of linear amplification, and production and fragmentation of biotinylated cRNA.

15µg of cRNA from each sample was hybridized to HG U133 Plus 2.0 microarrays.

Hybridization and scanning were performed according to the manufacture’s instruction

(Affymetrix). This step was performed at the PAN center of Stanford University. Data were normalized by GC robust multi-array average method and analyzed on

R/Bioconductor. BM2770 GMP, BM 2759 L-MPP, BM2761 CMP, BM2770 CMP were removed from further analysis due to low quality (GEO GSE6327).

147 a

HSC CD90+ CD45RA- Lin- CD34+CD38- MPP CD90- CD45RA- L-MPP CD90- CD45RA+ CMP CD123+ CLP Lin- CD34+CD38+ CD45R CD10+ MEP GMP CD123- CD123+ CD45RA- CD45RA+

Lin+ Erythrocyte Platelet Monocyte Granulocyte NKT B macrophage

b Figure 3.1. Schematic of human hematopoiesis with the immunophenotype of individual HSPC populations as indicated. Note the color scheme for each HSPC population is used throughout.

148 a

HSC CD90+ CD45RA- Lin- CD34+CD38- MPP CD90- CD45RA- L-MPP CD90- CD45RA+ CMP CD123+ CLP Lin- CD34+CD38+ CD45R CD10+ MEP GMP CD123- CD123+ CD45RA- CD45RA+

Lin+ Erythrocyte Platelet Monocyte Granulocyte NKT B macrophage

b

Figure 3.2. Pre-sort and post-sort FACS analysis of HSPCs from human bone marrow. Top panel: FACS-sorting scheme of six populations of HSPCs from normal human BM. Other panels: The second round of post-sort analysis to check the purity of sorting.

149 c Multidimensional Scaling

GMPGMP

GMP GMP

CMP CMP MEP CMP CMP MEP MEP Coordinate2 CMP MEPMEP

MPPMPP MPP HSCHSCMPP HSCHSC HSC

Coordinate1

d HMHB1 MIR539

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Figure 3.3. Comprehensive● ● ● DNA methylation analysis● shows tight clustering● of● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● HSPCs by their lineages.● Multidimensional● ● scaling examining the top 1,000 ● most ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● Beta ● variable methylation positions among normal● progenitors shows tight clustering of ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● distinct lineages. ● ● ● ● 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

0.1 0.1

0 0 + +MIR381 HMHB1 MIR487B MIR539 MIR889 Genes Genes chr5:143191200 143191600 143192000 143192400 chr14: 101513000 101513500 101514000 101514500

● ● ● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● MPP ● ● ● ● 80 ● 80 ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● CMP ● CMP ● ● ● ● ● 60 ● ● ● ● 60 ● ● ● ● ●

● ● GMP GMP ● ● ● ● ● ● ● ●

● ●

● 40 MEP 40 ●● ● MEP

● ● ● ●

● ● ● ● Methylated Methylated ● ● ● ●

20 ● 20 ● ● ● methylation methylation Non- Non- ●

● ● 0 ● methylated 0 methylated CG#: 1 2 3 4 CG#: 1 2 3 4 5 6

Supplementary Figure 6. Comprehensive DNA methylation analysis shows tight clustering of human hematopoietic stem and progenitor cells (HSPCs) (a) Schematic of human hematopoiesis with the immunophenotype of individual HSPC populations as indicated.150 Note the color scheme for each HSPC population is used throughout. (b) Pre-sort and post-sort FACS analysis of HSPCs from human bone marrow. Top panel: FACS-sorting scheme of six populations of HSPCs from normal human BM. Other panels: The second round of post-sort analysis to check the purity of sorting. (c) Multidimensional scaling examining the top 1,000 most variable methylation positions among normal progenitors shows tight clustering of distinct lineages. (d) DMR plots indicating genomic loci for newly identified genes with previously unknown functions in hematopoiesis HMHB1 and MIR539. Top: level of CpG methylation (beta) of each sample for the region; Middle: CpG density (curve), CpG sites (black tick marks), CpG islands (red lines); Bottom: gene annotation; Lower panel: bisulfite pyrosequencing replicating the methylation value for individual CpGs in the red boxes.

MPO CDK6Location

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● ● ● ● ● ● Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ●

● 0.00.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

0.1 0.1 CpG density 0.0 0.0 + + MPO CDK6 Genes location on chr17 Genes Genes chr17:56354600 56354800 56355000 56355200 56355400 chr7: 92237600 92238000 92238400 Genes

● ● ● ● ● ● ●● ●●● ●● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● CMP ● ● ● ● ● ● ● ● ● ● CMP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 60 ● ● ● ● ● ● ● ● ● ● ● 60 ● ● ● ● ● ● ● ● ● ● GMP ● ● ● ● ● ● ● ●● ● ● ● GMP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MEP ● ● ● ● MEP ● ● 40 ● ● ●● ● ● 40 ● ● ● ● ● ● ● ● ● ● ● ● Methylated Methylated ●

●

● 20 ● ● ● 20 Non- methylation ● Non- methylation ● ● ● ● ● ● ● 0 methylated 0 methylated CG#: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 CG#: 1 2 3 4 5 6 7 8 9 10 11 12

Figure 3.4. DMRHMHB1 plots indicating genomic loci for genes MIR539 with previously known

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● functions in hematopoiesis MPO and CDK6. Top:● ● level of CpG● methylation (beta) of ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● Beta ● ● ● ● ● each sample ● for the region; Middle: CpG density (curve), CpG ●sites (black tick marks), ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 0.2 0.4 0.6 0.8 1.0 CpG islands (red lines); Bottom: gene annotation;0.0 0.2 0.4 0.6 0.8Lower 1.0 panel: bisulfite pyrosequencing 0.1 0.1

0replicating the methylation value for individual 0CpGs in the red boxes. + +MIR381 HMHB1 MIR487B MIR539 MIR889 Genes Genes chr5:143191200 143191600 143192000 143192400 chr14: 101513000 101513500 101514000 101514500

● ● ● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● MPP ● ● ● ● 80 ● 80 ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● CMP ● CMP ● ● ● ● ● 60 ● ● ● ● 60 ● ● ● ● ●

● ● GMP GMP ● ● ● ● ● ● ● ●

● ●

● 40 MEP 40 ●● ● MEP

● ● ● ●

● ● ● ● Methylated Methylated ● ● ● ●

20 ● 20 ● ● ● methylation methylation Non- Non- ●

● ● 0 ● methylated 0 methylated CG#: 1 2 3 4 CG#: 1 2 3 4 5 6

151 MPO CDK6Location

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● ● ● ● ● ● Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ●

● 0.00.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

0.1 0.1 CpG density 0.0 0.0 + + MPO CDK6 Genes location on chr17 Genes Genes chr17:56354600 56354800 56355000 56355200 56355400 chr7: 92237600 92238000 92238400 Genes

● ● ● ● ● ● ●● ●●● ●● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● CMP ● ● ● ● ● ● ● ● ● ● CMP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 60 ● ● ● ● ● ● ● ● ● ● ● 60 ● ● ● ● ● ● ● ● ● ● GMP ● ● ● ● ● ● ● ●● ● ● ● GMP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MEP ● ● ● ● MEP ● ● 40 ● ● ●● ● ● 40 ● ● ● ● ● ● ● ● ● ● ● ● Methylated Methylated ●

●

● 20 ● ● ● 20 Non- methylation ● Non- methylation ● ● ● ● ● ● ● 0 methylated 0 methylated CG#: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 CG#: 1 2 3 4 5 6 7 8 9 10 11 12

HMHB1 MIR539

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

0.1 0.1

0 0 + +MIR381 HMHB1 MIR487B MIR539 MIR889 Genes Genes chr5:143191200 143191600 143192000 143192400 chr14: 101513000 101513500 101514000 101514500

● ● ● ● ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● MPP ● ● ● ● 80 ● 80 ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● CMP ● CMP ● ● ● ● ● 60 ● ● ● ● 60 ● ● ● ● ●

● ● GMP GMP ● ● ● ● ● ● ● ●

● ●

● 40 MEP 40 ●● ● MEP

● ● ● ●

● ● ● ● Methylated Methylated ● ● ● ●

20 ● 20 ● ● ● methylation methylation Non- Non- ●

● ● 0 ● methylated 0 methylated CG#: 1 2 3 4 CG#: 1 2 3 4 5 6

Figure 3.5. DMR plots indicating genomic loci for newly identified genes with

previously unknown functions in hematopoiesis HMHB1 and MIR539. Top: level of

CpG methylation (beta) of each sample for the region; Middle: CpG density (curve), CpG

sites (black tick marks), CpG islands (red lines); Bottom: gene annotation; Lower panel:

bisulfite pyrosequencing replicating the methylation value for individual CpGs in the red

boxes.

152 a HSC vs GMP

Overlap with CpG island c P<0.01 * 0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3 Proportion 0.2 P=0.03 0.2 *

0.1 Median proportion 0.1

0.0 0.0 Island Island Overlap Overlap Shore Shore Open sea Open sea Shelf Shelf b HSC vs MEP

Overlap with CpG island c P<0.01 *

0.5 0.5

0.4 0.4

0.3 0.3 P<0.01 *

Proportion 0.2 P<0.01 0.2 * 0.1 0.1 Median proportion

0.0 0.0 Island Island Overlap Overlap Shore Shore Open sea Open sea

Figure 3.6. Location of DMRs Shelf for normal hematopoiesis Shelf relative to CpG island. Left panel shows proportion of DMRs located in CpG island, overlapping region (50% overlap with island), shore, shelf, and open sea. Right panel shows distribution of length- matched random regions relative to CpG island, overlap, shore, shelf, and open sea. (a)

HSC vs GMP. (b) HSC vs MEP.

153 a Island Shore Differential expression Differential expression

Differential methylation Differential methylation Shelf Open sea Differential expression Differential expression

Differential methylation Differential methylation b Island Shore Differential expression Differential expression

Differential methylation Differential methylation Shelf Open sea Differential expression Differential expression

Differential methylation Differential methylation

154 Figure 3.7. Gene expression inversely correlates with DMRs at non-CpG island regions in normal hematopoiesis. DMRs located within 2kb of gene TSSs (black dots) were classified into 4 groups according to the distance relative to CpG island: island, shore, shelf, and open sea. DMRs located further than 2kb of gene TSSs are denoted as black pluses in the middle. Log2 ratios of differential expression were plotted against differential methylation (all values are from group2-group1). Wilcoxon rank-sum test was performed to test the null hypothesis that the expression differences for the hypo- or hypermethylated DMRs within 2kb of gene TSSs (black dots) showed stronger inverse correlation than the expression differences of the random DMRs that are located further than 2kb of TSSs (black pluses). Random DMRs were shown in the middles of DNA methylation axis regardless of their methylation differences. (a) For HSC vs GMP, shore showed statistically inverse correlation of DMR with gene expression. (b) For HSC vs

MEP, shore and open sea showed statistically inverse correlation of DMR with gene expression.

155 a b ZNF217 TREM1

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ●

● Beta Beta

● ● ● ● ● ● ● ● ● ● ● HSC ● HSC ● ● ● ● ● MPP ● ● MPP ● ● ● ● ● ● ● ● CMP ● CMP ● GMP GMP MEP MEP 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

0.1 0.1

0.0 0.0 + + ZNF217 TREM1 Genes Genes

chr20: 52199550 52199600 52199650 52199700 52199750 chr6: 41253800 41254200 41254600 41255000

Figure 3.8. Examples of DMRs located in master transcription factor or super-

enhancer. (a) ZNF217, a master transcription factor in monocyte differentiation. (b)

TREM1, located in monocyte specific super-enhancer.

156 a b Normal Hematopoiesis CpG methylation HSC CD90+ CD45RA- GMP GMP

Lin- CD34+CD38- MPP GMP CD90- GMP CD45RA-

L-MPP CMP CD90- CMP CD45RA+ MEP CMP CMP MEP Coordinate2 MEP CMP CMP MEPMEP CD123+ CLP CD45RA- CD10+ Lin- CD34+CD38+ MPPMPP MEP GMP HSC MPP CD123- CD123+ HSCHSC MPP HSC CD45RA- CD45RA+ HSC CMP GMP HSC MEP MPP

Coordinate1 Erythrocyte Platelet Monocyte/ Granulocyte T cells NK cells B cells Lin+ macrophage

c

inside exon : MPO inside exonLocation : CDK6

● ● ● ● ● ● ● ● ● ● ● 1.0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

0.6 0.8 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● ● ● ● Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

0.4 ● ● ● ● ● ●

● ● ● ● ●

● 0.0 0.2 0.0 0.2 0.4 0.6 0.8 1.0 0.15 0.15

0.10 0.10

CpG density 0.05 0.05 CpG density CpG density 0.00 0.00

+ + Genes Genes Genes MPO location on chr17 CDK6 chr17: 56354600 56354800 56355000 56355200 56355400 chr7: 92237400 92237600 92237800 92238000 92238200 92238400 Genes

● ● ● ● ● ● ●● ●●● ●● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● HSC ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● MPP ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● CMP ● ● ● ● ● ● ● ● CMP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 60 ● ● ● GMP ● ● ● ● GMP ● 60 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MEP ● MEP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● methylation 40 Methylated ● ● ● Methylated ● ● 40 ● ●

methylation ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● Non- Non- ● ● ● ●

● 20 ● 20 ● methylated methylated ● ● ●

● ●

● ● ● ● ● ● ● 0 0 CG#: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 CG#: 1 2 3 4 5 6 7 8 9 10 11 12

d HMHB1 covers : MIR539

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Beta ● ● ● Beta ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.0 0.15 0.15 0.10 0.10 0.05 0.05 CpG density

CpG density 0.00 0.00 MIR381+ MIR487B MIR539 MIR889 + HMHB1

Genes Genes chr14: 101513000 101513500 101514000 101514500 chr5:143191200 143191400 143191600 143191800 143192000 143192200 143192400 143192600

● ●

● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● 100 ● ● ● ● ● ● ● ● ● ● ● HSC ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● MPP ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 80 ● ● ●

● ● 80 ● ● ● ● ●

● ● ● ● ● CMP ● ● ● ● ● CMP ● ● ● ● ● ● ● ● ● ● 60 ● ● ● ● GMP ● ● ● 60 ● ● GMP ● ● ●

● ● ● ● ● MEP ● ● ● ● MEP ● ● ● methylation ● ● ● ● ● 40 Methylated

methylation 40 ● Methylated ● ●

● ● ● Non-

● ● ● Non- ● ● 20 ● 20 ● methylated ● ● methylated ● ●

● ● ● 0 0 CG#: 1 2 3 4 5 6 CG#: 1 2 3 4

L-MPP Mouse H e yp Human om eth yla tion GMP

on lati thy e e om ng yp ha CMP H l c HSC MPP oba gl No Hypomethylaiton H ypo Hypermethylation me thy lat MEP ion

Figure 3.9. Global methylation changes during hematopoietic development in

human and mouse. The direction of global methylation change is shown for each

comparison in red for human and blue for mouse hematopoiesis. The dotted lines

represent two comparisons shared between the studies for human and mouse

hematopoiesis.

157 Table 3.1. Normal bone marrow donor sample analysis

Sample ID Age Gender Application

BM2627 30 M 450K BM2710 29 F 450K BM2712 39 M 450K BM2748 24 M 450K BM2753 22 F 450K BM2759 38 M GEP BM2761 26 M GEP BM2768 25 M GEP BM2770 39 F GEP BM2793 18 F GEP BM2794 21 M GEP BM2806 35 M GEP BM3604 26 M P BM3668 24 M P BM3671 24 M P

Abbreviations: 450K, Illumina Infinium Human Methylation 450K BeadChip array; F, female; GEP, gene expression profiling microarray; M, male; P, bisulfite pyrosequencing

158 Table 3.2. Antibodies for flow cytometry

Cell Surface Catalog Working Fluorophore Manufacturer Application marker Number Dilution CD2 555328 CD3 555341 CD4 555348 CD7 555362 CD8 555368 CD10 555376 PE-Cy5 1:50 CD11b 555389 CD14 340585 BD To sort normal CD19 555414 Bioscience HSPCs from BMs CD20 555624 CD56 555517 CD235a 559944 CD34 APC 340667 1:50 CD38 PE-Cy7 335790 1:100 CD45RA PB 560362 1:25 CD90 FITC 555595 1:25 CD123 PE 554529 1:25

159 Table 3.3. DMR lists for pair-wise comparisons among HSPCs (See Appendix2)

160 Table 3.4. Summary of DMRs identified in the indicated pairwise comparisons

Comparisons Numbers of DMRs* Locations of DMRs relative to CpG islands (%)

(Group1 versus Group2) Group1>Group2 Group1

MPP vs CMP 158 8 3.6 22.3 10.2 63.9

CMP vs GMP 366 362 2.5 17.0 15.4 65.1

CMP vs MEP 319 13 3.6 25.6 16.0 54.8

GMP vs MEP 1308 764 2.4 19.3 15.3 63

MPP vs L-MPP 49 54 2.9 15.5 19.4 62.1

HSC vs L-MPP 165 109 2.2 19.0 15.0 63.7 L-MPP vs GMP 556 0 2.2 15.5 13.3 69.1

HSC vs GMP 1168 162 2.0 17.8 13.5 66.7

HSC vs MEP 1545 190 2.5 23.7 14.9 58.9

161 Table 3.5. Enrichment of DMRs for normal hematopoiesis in super-enhancers

Tissues or Cell lines Cell type DMR lists* HSC vs GMP HSC vs MEP CD34+(adult) HSPC 0.009 2.5 x 10-9 CD14 Monocyte 6.9 x 10-16 0.12 Jurkat T cell leukemia 5.3 x 10-6 1 K562 Erythroleukemia(CML) 1 9.6 x 10-9 MM1S Multiple myeloma 0.04 0.08 Adipose Adipose 1 1 Angular gyrus Brain 0.15 0.66 H1 Embryonic stem cell 0.90 0.78 Spleen Spleen 1.1 x 10-5 0.003 * Family wise error rate (FWER) cutoff of 0.1 was used to select DMRs

162 Table 3.6. DMRs of normal hematopoiesis located in super-enhancer of different tissues and cell types (See Appendix3)

163 Table 3.7. Common genes between mouse and human hematopoiesis (See Appendix4)

Columns are chromosome, start, end, diffMethyl (difference of methylation percentage for group2-group1 for group1 vs group2 comparison), p.value, fwer, Gene, annotation, relation to gene, distance to gene.

Yellow colored DMRs are showing opposite direction of methylation change compared to mouse hematopoietic DMRs.

164 Table 3.8. Primers for bisulfite pyrosequencing

Gene Primer type Sequence(5'→3')

MIR539 Nested forward TATGATAAGTTTTGTAAAGGGATGTA

Nested reverse /5Biosg/CAAATCCCTAATAACACCAAAAAAT

Long forward GTGTTGTTGTTTTATATTTGAGGAGAA

Long reverse CATATCCAAAAAATACCTCCAAAAA

Sequencing 1 (F) TGATAAGTTTTGTAAAGGGATG

Sequencing 2 (F) GTTTTAATTTTAGAATTTTGGA

CDK6 Nested forward TGTTTTTGAGATAGTAGTAGGGTATTTTG

Nested reverse /5Biosg/TAACCAATCTAAACCCCATTTACTC

Long forward GGGGTAGATAGTTTTATATAGGGTAGTTGT

Long reverse TTCCACCCCAAAATTTATTATAACA

Sequencing 1 (F) GATAGTAGTAGGGTATTTTGAT

Sequencing 2 (F) ATTGTTTTTTTTTTGTTAAAGG

Sequencing 3 (F) TAAGTGGGAATTAAGTTTTGAG

SLC39A4 Nested forward AGGGGAAGGTAGATTTTAGGGTAG

Nested reverse /5Biosg/AACCCCAAAACACTAAACTCAATAC

Long forward TTTTGAGTTTAGAGGTTTTATTTTTAT

Long reverse ACAAACTCCCCTAAAAACCC

Sequencing 1 (F) AGGGGGAGTTTAGATGTTATTT

Sequencing 2 (F) GTTTGAGGTTTAAGGATTTTGT

ZDHHC14 Nested forward GGGAAAGAAGAGAATTATTTTTAGGTT

Nested reverse /5Biosg/CAAACCCAATACCTCTATCAAAATC

Long forward GTGTTAATGGTATTTTTTGATAGT

Long reverse AAACTATCTTTACTTTTACTCAAAC

Sequencing 1 (F) GAGGAAATGGGAGTTTTGTTTT

Sequencing 2 (F) GAGTTTATTTATGTTTGTAGAT

Sequencing 3 (F) AAATATATTTTTTTTTTTTATT

Sequencing 4 (F) GAAGTTTTTTTTGATTTTTGTTT

165 TRPM2 Nested forward GGGTTGTTTAGAAGGGTTATTGATT

Nested reverse /5Biosg/CCCAATTCTATTCTCCCAAAAATATA

Long forward TTTTTAGTTTTGAGGAAAGTTGGTT

Long reverse AATAATCATACAACCCACAAAAAAC

Sequencing 1 (F) GATTTGGGGATGGTTTTTAATT

Sequencing 2 (F) GAGTTGGAGGTTATAGTGTTTT

Sequencing 3 (F) TTGTTGGTTAGTTTGTAGTTGG

Sequencing 4 (F) TTGTTGGTTAGTTTGTAGTTGG

MAEA Nested forward TTAGTTTAGGATGGTAGGAAGTTAT

Nested reverse /5Biosg/TTATCTTTTACAATTAACCAAAAAA

Long forward TTTTTAAGAAGTTTTTTAGGGGATT

Long reverse ATATCATCATCATATTTTCACCAACAC

Sequencing 1 (F) TAGTTTAGGATGGTAGGAAGTT

Sequencing 2 (F) AATATGAGATTGGTTTTTTTAG

Sequencing 3 (F) TATTTTTGGTTTTGTGATGAGT

STAT3 Nested forward GGTTTTTAATGTAGGTAATTTGTTGT

Nested reverse /5Biosg/TATTTTAATTTCCAACCAAAACATC

Long forward TTTTTTATTTTTTATTAGTTTTTTAGT

Long reverse CTTACTTAATTTCTAAAAAATTCCTACTCT

Sequencing 1 (F) TTTTAATGTAGGTAATTTGTT

Sequencing 2 (F) GTGAGAGTTTTTTG

HMHB1 Nested forward TGGAGAAATTAGAATTGGAGGAGTA

Nested reverse /5Biosg/CTAAATAATCCCAACAACAAAAACC

Long forward ATGAGGAAATTATATTTTAGGAGGT

Long reverse CAACCAAACAATAAACTATAAAACC

Sequencing 1 (F) GAGAAGAAAAAAGAGGTGAGGG

Sequencing 2 (F) TATAATAGGTGAAAATAGGGAT

MPO Nested forward TAGTTTTAGTTGGTTGGATATGTTG

Nested reverse /5Biosg/AACCTCTCTCTATACCTCAAATCCC

Long forward TAGGTTGTTAAAGGGTAGTAGGGTT

166 Long reverse TACCAAAAATCCTAAAAACAAAAAA

Sequencing 1 (F) AGTTTTAGTTGGTTGGATATGT

Sequencing 2 (F) GTAGGTTTTTGGTTAGGGGTTT

Sequencing 3 (F) GGATGGTGATGTTGTT

/5Biosg/ = 5’ biotin added

F = forward

167 Chapter 4

The cell of origin of leukemia stem cell

168 This work is an ongoing project of the Feinberg Lab and Johns Hopkins. All publication

rights are reserved for these institutions and the presentation of this work here does not

preclude future publication elsewhere.

Summary

The question of the cell of origin of LSC has long been remained in the field, yet

much of it still needs to be understood. We hypothesized that DNA methylation could be

useful to infer the cell identity of LSCs compared to HSPCs, as it has been shown that transformed or reprogrammed cells retained an epigenetic memory of the cell of origin

(Kim et al., 2010; Kim et al., 2011; Polak et al., 2015). We were able to provide a clue for the cell of origin of LSCs by comparing the DNA methylation profile of LSCs to normal

HSPCs, obtained from previous chapter. Clustering analysis based on DNA methylation status showed LSCs clustered with either L-MPP or GMP. This result was replicated in

TCGA data set, a larger cohort, and supported by molecular characteristics of each cluster such as FAB types, genetic mutations, and cytogenetic risk groups. These results suggested L-MPP and GMP would be major sources of LSCs.

Results

LSC Forms Two Clusters with L-MPP-like and GMP-like

In order to relate normal hematopoiesis to LSC, we first identified the DMRs from all possible pairwise comparisons among the 6 HSPC after applying a more rigorous cutoff of FWER<0.1 (Table 4.1). The resulting 216 DMRs were applied in clustering analysis including all 6 normal HSPC populations with LSCs and Blasts

169 (Figure 4.1). Strikingly, this analysis revealed that AML samples formed 2 distinct

clusters, L-MPP-like and GMP-like (Figure 4.1). Importantly, the GMP-like cluster

included several CD34+CD38- subpopulations, indicating that these clusters could not

have been identified by immunophenotype alone. Moreover, clustering analysis using an

equal number of length-matched random regions showed that the clustering of AML

populations with either L-MPP or GMP was unique to the selected DMRs (Figure 4.2).

TCGA AML Samples Form Two Major Clusters, L-MPP-like and GMP-like

Strikingly, using the same 216 DMRs, the TCGA samples also formed the same

two major clusters, L-MPP-like and GMP-like (Figure 4.3). In addition to the two major

clusters, we also identified a minor CMP-like cluster that was not observed in our smaller

cohort. We calculated scores indicating the similarity of each TCGA sample to each of

the six progenitors, and designated a counterpart HSPC population for each TCGA sample based on highest similarity. This approach showed that 76.6% of TCGA samples resembled GMP and 14.6% had a methylation profile most similar to L-MPP (Figure 4.4).

We hypothesized that if the assignment of AML samples to L-MPP-like and

GMP-like clusters was related to the cell of origin, then the degree of maturity and

morphology might differ between the two groups. Consistent with this, we compared the

distribution of the French-American-British (FAB) classification of the TCGA samples

and found that the L-MPP-like cases mainly consisted of more immature M0, M1, and

M2 types, while the more differentiated M4 and M5 types were enriched in GMP-like

AML (p<1x10-4, chi-square test, Figure 4.5). It should be noted that the LSC epigenetic

signature is not merely a recapitulation of FAB types, as our signature is prognostic in

multivariate analysis while FAB types are not. In addition, it is not possible to know the

170 cell of origin simply by examining FAB types (Table 4.2). For example, in 38 cases of

M1 AML, 29/38 are GMP-like and 9/29 are L-MPP-like (Table 4.2).

Finally, we sought to investigate if the L-MPP-like and GMP-like clusters, and

therefore the potential cell of origin, were associated with cytogenetic abnormalities or

recurrent mutations of specific genes including DNMT3A, IDH1, IDH2, TET1, TET2,

FLT3, and NPM1. The GMP-like cluster was enriched for patients in the low and intermediate cytogenetic risk groups, while the L-MPP-like cluster was enriched for patients in the high cytogenetic risk group (p=1x10-4, Fisher’s exact test; Figure 4.6a).

We found that IDH1 and IDH2 mutations were enriched in the L-MPP like group (p<0.01 for both, Fisher’s exact test), and FLT3 and NPM1 mutations were enriched in the GMP- like group (p<0.01 for both, Fisher’s exact test). DNMT3A and TET1 mutations were more enriched in the L-MPP group, but this was not statistically significant (Figure 4.6b).

Together, these results demonstrate that DNA methylation signatures permit a novel clustering of AML into L-MPP-like and GMP-like groups that may reflect the cell of origin for each case and demonstrate an association with key disease features.

Discussion

As DNA methylation is a potential marker of cell identity, here we compared

DNA methylomes of normal HSPC to LSC as a marker for the cell of origin in AML.

Using this approach, we observed two subtypes in our cohort: L-MPP-like and GMP-like.

These two subtypes were also identified in the TCGA cohort, suggesting that leukemic transformation predominantly occurs at either the L-MPP or GMP stage of hematopoietic development. We found that other features of AML were associated with these two

171 subtypes including FAB type, several mutations, and cytogenetic abnormalities, suggesting that the cell of origin may drive key clinical features in AML.

We newly identified a small subset of AML cells clustering with CMP and few samples clustering with HSC, MPP, and MEP that could not be identified in smaller datasets including our own data and a previous study (Goardon et al., 2011). The result from TCGA indicates that the cell of origin of AML could be variable among HSPCs.

The result for the cell of origin should be considered carefully, as it is hard to provide definite proof of the question due to experimental limitations. Despite this caveat that affects all of human primary cancer biology, we do provide the first epigenetic evidence for cell of origin in human leukemia and believe that our approach using epigenomic profiles suggests an efficient way to study cell of origin in cancer biology using large data sets.

Materials and Methods

Illumina Infinium Human Methylation 450 Bead Array Assay

Genomic DNA from each sample was purified using the MasterPure DNA purification kit (Epicentre) according to the manufacturer’s protocol. The genomic DNA (250-500ng) was treated with sodium bisulfate using the Zymo EZ DNA Methylation Kit (ZYMO

Research) as recommended by the manufacturer, with the alternative incubation conditions for the Illumina Infinium Methylation Assay. Converted DNA was eluted in

11ul of elution buffer. DNA methylation level was measured using Illumina Infinium HD

Methylation Assay (Illumina) according to the manufacturer’s specifications.

172 Methylation array data are deposited at the Gene Expression Omnibus (GEO) with accession number GSE63409.

Illumina Infinium Human Methylation 450 Bead Array Analysis

Raw intensity files were obtained using minfi package (Aryee et al., 2014) to calculate methylation ratios (Beta values). The data was normalized using Illumina preprocessing method implemented in minfi. Several quality control measures were applied to remove arrays with low quality. Control probes were examined on the 450k array to assess several measures including bisulfite conversion, extension, hybridization, specificity and others. One of the MPP samples (BM2712) and two samples of TCGA (Patient ID: 2934 and 2827) showed low quality for the measures, so they were removed for further analysis. Next, median methylated and unmethylated signals were calculated for each arrays; no array was identified for signal values lower than 10.5. For multidimensional scaling analysis, probes containing an annotated SNP (dbSNP137) at the single-base extension or CpG sites were removed (17398 probes removed). Minfi 1.8.9 was used.

Bump hunting method previously described was applied to identify DMRs in 450k array

(Aryee et al., 2014; Jaffe et al., 2012). Beta value of 0.1 (10% of methylation difference) was used as cutoff when finding DMRs. Bumphunter 1.2.0 was used.

Statistical Analysis

To assign cell identity of normal HSPCs to TCGA samples, mean methylation value of each 216 DMRs for normal hematopoiesis (methylation profile) was retrieved and standard deviation of the mean value for each signature was calculated. Then scores

(probability density values as log value) for each TCGA sample regarding normal HSPCs’ profile was calculated using dnorm function with the mean and standard deviation

173 calculated in previous step. Maximum value of scores among the ones regarding normal

HSPC methylation profile was chosen, and then cell identity assigned.

For clustering analysis, hclust function with ward method in R was used to generate all the cluster dendrogram analysis.

174 Figure 4

a 0 10 20 30 40 50 60 BM2748_CMP BM2627_CMP SU266_CD34- BM2710_HSC BM2753_MPP BM2753_MEP BM2712_MEP BM2753_CMP BM2753_HSC BM2710_MPP BM2710_CMP BM2712_CMP BM2712_HSC BM2748_HSC BM2627_HSC BM2748_MPP BM2627_MPP SU001_CD34- SU032_CD34- BM2748_MEP BM2627_MEP BM2712_GMP BM2627_GMP SU056_CD34- BM2753_GMP SU302_CD34- BM2748_GMP BM2710_GMP SU036_CD34- BM2710 _MEP SU006_CD34- SU306_CD34- SU267_CD34- SU046_CD34- SU029_CD34- SU014_CD34- SU035_CD34- SU008_CD34- SU042_CD34- SU306_CD34+38- SU056_CD34+38+ SU056_CD34+38- SU046_CD34+38+ SU306_CD34+38+ SU035_CD34+38- SU008_CD34+38- SU001_CD34+38+ SU001_CD34+38- SU036_CD34+38+ SU036_CD34+38- SU032_CD34+38+ SU032_CD34+38- SU006_CD34+38+ SU006_CD34+38- SU035_CD34+38+ SU008_CD34+38- SU029_CD34+38+ SU029_CD34+38- SU302_CD34+38- SU302_CD34+38+ SU267_CD34+38- SU267_CD34+38+ SU266_CD34+38- SU266_CD34+38+ SU042_CD34+38+ SU042_CD34+38- SU014_CD34+38+ SU014_CD34+38- Figure 4.1. Epigenetic signatures define subgroups of AML LSC reflecting the cell b of origin. A total of 216 DMRs identified from all possible pairwise comparisons among HSC MPP 6 HSPCs were used to cluster all normal HSPC with all AML subpopulations. The L-MPP CMP primary AML subpopulations form two major clusters: L-MPP-like and GMP-like. LSC GMP MEP TCGA subpopulations are indicated in bold. 0 20 40 60 80 100

GMP-like cluster L-MPP-like cluster c d Cell type TCGA % L-MPP-like AML HSC 1 0.5 GMP-like AML MPP 3 1.6 L-MPP 28 14.6 CMP 12 6.3 GMP 147 76.6 MEP 1 0.5 192 100 FAB type(%)

0 10 20 30 40 50 M0M1 M2 M3 M4 M5 M6 M7 NA

175 15

10 5 0 SU001_CD34- SU056_CD34- BM2748_MEP BM2712_MEP BM2710_HSC BM2627_GMP SU046_CD34- BM2710_MPP BM2627_MEP BM2710_CMP BM2753_HSC SU306_CD34- BM2753_MEP BM2753_MPP BM2712_CMP SU267_CD34- BM2753_CMP BM2710_GMP BM2712_HSC BM2712_GMP BM2627_CMP BM2748_GMP SU032_CD34- BM2748_HSC BM2748_CMP SU036_CD34- BM2627_HSC BM2748_MPP BM2753_GMP SU266_CD34- BM2710 _MEP SU014_CD34- BM2627_MPP SU035_CD34- SU302_CD34- SU006_CD34- SU042_CD34- SU029_CD34- SU008_CD34- SU306_CD34+38- SU006_CD34+38- SU046_CD34+38+ SU266_CD34+38- SU036_CD34+38- SU306_CD34+38+ SU032_CD34+38+ SU032_CD34+38- SU001_CD34+38+ SU001_CD34+38- SU036_CD34+38+ SU266_CD34+38+ SU029_CD34+38+ SU056_CD34+38- SU008_CD34+38- SU035_CD34+38+ SU035_CD34+38- SU267_CD34+38- SU267_CD34+38+ SU006_CD34+38+ SU008_CD34+38+ SU056_CD34+38+ SU014_CD34+38+ SU014_CD34+38- SU042_CD34+38+ SU042_CD34+38- SU029_CD34+38- SU302_CD34+38- SU302_CD34+38+

Figure 4.2. Clustering analysis of AML populations with normal HSPCs using Supplementary Figure 8. Clustering analysis of AML populations with normal lengthHSPCs matched using lengthrandom matched 216 regions random. Clustering 216 regions analysis. Clustering using an randomalysis using length random matched length matched regions shows no clustering between AML populations or normal HSPCs. Normal progenitors clustered together, but not by lineages. regions shows no clustering between AML populations or normal HSPCs. Normal progenitors clustered together, but not by lineages.

176 Figure 4

a 0 10 20 30 40 50 60 BM2748_CMP BM2627_CMP SU266_CD34- BM2710_HSC BM2753_MPP BM2753_MEP BM2712_MEP BM2753_CMP BM2753_HSC BM2710_MPP BM2710_CMP BM2712_CMP BM2712_HSC BM2748_HSC BM2627_HSC BM2748_MPP BM2627_MPP SU001_CD34- SU032_CD34- BM2748_MEP BM2627_MEP SU056_CD34- BM2712_GMP BM2627_GMP BM2753_GMP SU302_CD34- BM2748_GMP BM2710_GMP SU036_CD34- BM2710 _MEP SU006_CD34- SU306_CD34- SU267_CD34- SU046_CD34- SU029_CD34- SU014_CD34- SU035_CD34- SU008_CD34- SU042_CD34- SU306_CD34+38- SU056_CD34+38+ SU056_CD34+38- SU046_CD34+38+ SU306_CD34+38+ SU035_CD34+38- SU008_CD34+38- SU001_CD34+38+ SU001_CD34+38- SU036_CD34+38+ SU036_CD34+38- SU032_CD34+38+ SU032_CD34+38- SU006_CD34+38+ SU006_CD34+38- SU035_CD34+38+ SU008_CD34+38- SU029_CD34+38+ SU029_CD34+38- SU302_CD34+38- SU302_CD34+38+ SU267_CD34+38- SU267_CD34+38+ SU266_CD34+38- SU266_CD34+38+ SU042_CD34+38+ SU042_CD34+38- SU014_CD34+38+ SU014_CD34+38-

b HSC MPP L-MPP CMP GMP MEP TCGA 0 20 40 60 80 100

GMP-like cluster L-MPP-like cluster c d Cell type TCGA % L-MPP-like AML HSC 1 0.5 GMP-like AML MPP Figure3 4.3.1.6 Epigenetic signatures define subgroups of AML samples in TCGA L-MPP 28 14.6 CMP 12 6.3 GMP reflecting147 the76.6 cell of origin. Clustering analysis of TCGA AML samples with normal MEP 1 0.5 human192 HSPC100 using the 216 DMRs shows that the L-MPP-like and GMP-like clusters are FAB type(%) observed in this cohort as well. 0 10 20 30 40 50 M0M1 M2 M3 M4 M5 M6 M7 NA

177 Figure 4

a 0 10 20 30 40 50 60 BM2748_CMP BM2627_CMP SU266_CD34- BM2710_HSC BM2753_MPP BM2753_MEP BM2712_MEP BM2753_CMP BM2753_HSC BM2710_MPP BM2710_CMP BM2712_CMP BM2712_HSC BM2748_HSC BM2627_HSC BM2748_MPP BM2627_MPP SU001_CD34- SU032_CD34- BM2748_MEP BM2627_MEP SU056_CD34- BM2712_GMP BM2627_GMP BM2753_GMP SU302_CD34- BM2748_GMP BM2710_GMP SU036_CD34- BM2710 _MEP SU006_CD34- SU306_CD34- SU267_CD34- SU046_CD34- SU029_CD34- SU014_CD34- SU035_CD34- SU008_CD34- SU042_CD34- SU306_CD34+38- SU056_CD34+38+ SU056_CD34+38- SU046_CD34+38+ SU306_CD34+38+ SU035_CD34+38- SU008_CD34+38- SU001_CD34+38+ SU001_CD34+38- SU032_CD34+38+ SU032_CD34+38- SU036_CD34+38+ SU036_CD34+38- SU006_CD34+38+ SU006_CD34+38- SU035_CD34+38+ SU008_CD34+38- SU029_CD34+38+ SU029_CD34+38- SU302_CD34+38- SU302_CD34+38+ SU267_CD34+38- SU267_CD34+38+ SU266_CD34+38- SU266_CD34+38+ SU042_CD34+38+ SU042_CD34+38- SU014_CD34+38+ SU014_CD34+38-

b HSC MPP L-MPP CMP GMP MEP TCGA 0 20 40 60 80 100

GMP-like cluster L-MPP-like cluster c d Cell type TCGA % L-MPP-like AML HSC 1 0.5 GMP-like AML MPP 3 1.6 L-MPP 28 14.6 CMP 12 6.3 GMP 147 76.6 MEP 1 0.5 192 100

FAB type(%)

Figure 4.4. Cell identity of TCGA AML samples. TCGA samples were classified into 0 10 20 30 40 50 one of HSPC populations according to their DNA methylation profile M0 by generatingM1 M2 M3 M4 M5 M6 M7 NA methylation profiles of all the normal HSPC, and then calculating scores of each sample based on the closest population. The normal progenitor cell identity for each TCGA samples is summarized in this table.

178 Figure 4 a 0 10 20 30 40 50 60 BM2748_CMP BM2627_CMP SU266_CD34- BM2710_HSC BM2753_MPP BM2753_MEP BM2712_MEP BM2753_CMP BM2753_HSC BM2710_MPP BM2710_CMP BM2712_CMP BM2712_HSC BM2748_HSC BM2627_HSC BM2748_MPP BM2627_MPP SU001_CD34- SU032_CD34- BM2748_MEP BM2627_MEP BM2712_GMP BM2627_GMP SU056_CD34- BM2753_GMP SU302_CD34- BM2748_GMP BM2710_GMP SU036_CD34- BM2710 _MEP SU006_CD34- SU306_CD34- SU267_CD34- SU046_CD34- SU029_CD34- SU014_CD34- SU035_CD34- SU008_CD34- SU042_CD34- SU306_CD34+38- SU056_CD34+38+ SU056_CD34+38- SU046_CD34+38+ SU306_CD34+38+ SU035_CD34+38- SU008_CD34+38- SU001_CD34+38+ SU001_CD34+38- SU032_CD34+38+ SU032_CD34+38- SU036_CD34+38+ SU036_CD34+38- SU006_CD34+38+ SU006_CD34+38- SU035_CD34+38+ SU008_CD34+38- SU029_CD34+38+ SU029_CD34+38- SU302_CD34+38- SU302_CD34+38+ SU267_CD34+38- SU267_CD34+38+ SU266_CD34+38- SU266_CD34+38+ SU042_CD34+38+ SU042_CD34+38- SU014_CD34+38+ SU014_CD34+38- b HSC MPP L-MPP CMP GMP MEP TCGA 0 20 40 60 80 100

GMP-like cluster L-MPP-like cluster c d Cell type TCGA % L-MPP-like AML HSC 1 0.5 GMP-like AML MPP 3 1.6 L-MPP 28 14.6 CMP 12 6.3 GMP 147 76.6 MEP 1 0.5 192 100 FAB type(%) 0 10 20 30 40 50 M0M1 M2 M3 M4 M5 M6 M7 NA

Figure 4.5. Distribution of FAB types for L-MPP-like and GMP-like TCGA AML

samples. The L-MPP-like and GMP-like TCGA samples were grouped according to their

FAB classification. NA: not classified.

179 a Cell identity Favorable Intermediate/Normal Poor GMP-like 33 90 22 L-MPP-like 0 13 15

b L-MPP (%) GMP (%) DNMT3A 33.3 25.0 IDH1 33.3 6.3 IDH2 29.6 5.6 TET1 3.7 0.7 TET2 7.4 7.6 NPM1 3.7 34.7 FLT3 7.4 32.6

Figure 4.6. Correlation of disease features with cell identity. (a) Number of TCGA Supplementary Figure 9. Correlation of disease features with cell identity. (a) samplesNumber that of TCGA belong samples to each that cytogenetic belong to eachrisk cytogeneticgroup is shown risk group for GMP-like is shown for and GMP- L-MPP- like and L-MPP-like AML samples. Information on cytogenetic risk group was retrieved from clinical annotation of TCGA patients. (b) Percentage of AML samples with specific like AML samples. Information on cytogenetic risk group was retrieved from clinical mutations including DNMT3A, IDH1, IDH2, TET1, TET2, NPM1, and FLT3 are shown for L-MPP-like and GMP-like AML samples. Mutations that showed statistical significance of annotationassociation of withTCGA specific patients. cell types(b) Percentage are colored of in AML red. Information samples with on mutationsspecific mutations was retrieved from TCGA Mutation Annotation Format (MAF) file. including DNMT3A, IDH1, IDH2, TET1, TET2, NPM1, and FLT3 are shown for L-MPP- like and GMP-like AML samples. Mutations that showed statistical significance of association with specific cell types are colored in red. Information on mutations was retrieved from TCGA Mutation Annotation Format (MAF) file.

180 Table 4.1. DMRs for normal hematopoiesis

chr start end gene name

chr1 2164542 2164602 SKI

chr1 12538341 12538541 VPS13D

chr1 19652788 19652788 PQLC2

chr1 25598878 25599066 RHD

chr1 25747376 25747437 RHCE

chr1 39513326 39513326 NDUFS5

chr1 43290834 43291209 ERMAP

chr1 44114346 44114355 KDM4A

chr1 45103896 45103967 RNF220

chr1 51986130 51986254 EPS15

chr1 53103294 53103294 FAM159A

chr1 116193783 116193783 VANGL1

chr1 145828001 145828006 GPR89A

chr1 159866558 159866558 CCDC19

chr1 179016233 179016233 FAM20B

chr1 182760143 182761262 NPL

chr1 185253495 185253495 SWT1

chr1 199827580 199827580 NR5A2

chr1 226012913 226013010 EPHX1

chr1 226036279 226036279 TMEM63A

chr10 11631072 11631072 USP6NL

chr10 32621403 32621403 EPC1

chr10 70805603 70805603 KIAA1279

chr10 97036527 97036527 PDLIM1

chr10 100220809 100220809 HPSE2

chr10 135059097 135059289 MIR202

181 chr11 1297066 1297087 TOLLIP chr11 1325718 1325852 TOLLIP chr11 5276490 5276490 HBG2 chr11 8385712 8385767 STK33 chr11 12136405 12136405 MICAL2 chr11 33759043 33759043 CD59 chr11 47885166 47885166 NUP160 chr11 59823993 59824161 MS4A3 chr11 67251677 67251939 AIP chr11 69061454 69061473 MYEOV chr11 73681281 73681281 DNAJB13 chr11 108422791 108423124 EXPH5 chr11 117695591 117696015 FXYD2 chr12 1922058 1922067 CACNA2D4 chr12 14996508 14996587 ART4 chr12 25103173 25103643 BCAT1 chr12 32530696 32530696 BICD1 chr12 53602179 53602179 ITGB7 chr12 89938002 89938002 POC1B-GALNT4 chr12 96390059 96390143 HAL chr12 114232702 114232905 RBM19 chr12 117480333 117480333 TESC chr12 123632825 123633306 PITPNM2 chr13 41631052 41631052 WBP4 chr13 49147573 49147573 RCBTB2 chr13 113305704 113305901 C13orf35 chr13 114828264 114828455 RASA3 chr13 114918702 114918702 RASA3 chr14 21359295 21359943 RNASE3

182 chr14 81425912 81426015 TSHR chr14 101513572 101514051 MIR539 chr14 103565867 103566172 EXOC3L4 chr14 104190678 104190829 ZFYVE21 chr14 104197160 104197160 ZFYVE21 chr14 104625249 104625669 KIF26A chr15 60907749 60907749 RORA chr15 65066231 65066710 RBPMS2 chr15 80938098 80938098 ARNT2 chr15 90643766 90643766 IDH2 chr15 90727560 90727570 SEMA4B chr15 101720914 101720914 CHSY1 chr15 101777761 101777800 CHSY1 chr16 23880664 23880664 PRKCB chr16 56892460 56892460 MIR138-2 chr16 67567158 67567158 FAM65A chr16 85551478 85551748 KIAA0182 chr16 85622205 85622276 KIAA0182 chr16 85708039 85708039 KIAA0182 chr16 86011615 86012085 IRF8 chr16 88558065 88558237 ZFPM1 chr16 88563934 88564934 ZFPM1 chr17 998432 998504 ABR chr17 5138259 5138696 LOC100130950 chr17 14222235 14222235 HS3ST3B1 chr17 26517057 26517057 NLK chr17 38717206 38717275 CCR7 chr17 40489569 40489785 STAT3 chr17 45286354 45286609 MYL4

183 chr17 46648525 46648582 HOXB3 chr17 47372126 47372126 ZNF652 chr17 74844962 74845007 MGAT5B chr17 76142333 76142461 C17orf99 chr17 76850256 76850277 TIMP2 chr17 79623603 79623661 PDE6G chr17 80273242 80273322 CD7 chr17 80581701 80581925 WDR45L chr18 55250568 55250568 FECH chr18 60646614 60646671 PHLPP1 chr19 827715 827843 AZU1 chr19 2865997 2865997 ZNF556 chr19 12895288 12895529 JUNB chr19 12997997 12998927 KLF1 chr19 15391832 15391946 BRD4 chr19 39048014 39048014 RYR1 chr19 47839132 47839187 GPR77 chr19 50848024 50848461 NR1H2 chr19 51858208 51858276 ETFB chr19 52076691 52076691 ZNF175 chr19 55417390 55417647 NCR1

chr2 24233923 24234017 MFSD2B

chr2 37642493 37642493 QPCT

chr2 39355435 39355435 SOS1

chr2 55642894 55642894 CCDC88A

chr2 61407270 61407414 AHSA2

chr2 65593761 65594021 SPRED2

chr2 68348796 68348796 WDR92

chr2 74753691 74753759 AUP1

184 chr2 75089669 75089669 HK2

chr2 109789810 109789810 SH3RF3

chr2 113885116 113885277 IL1RN

chr2 128052778 128052889 ERCC3

chr2 144992352 144992352 GTDC1

chr2 149259371 149259371 MBD5

chr2 149639612 149639914 KIF5C

chr2 170779769 170779769 UBR3

chr2 172377981 172378036 CYBRD1

chr2 179149472 179149605 OSBPL6

chr2 220436894 220436948 INHA

chr2 239330171 239330202 ASB1

chr2 240225062 240225142 HDAC4 chr20 45319350 45319455 TP53RK chr20 52199594 52199778 ZNF217 chr21 43823749 43824262 UBASH3A chr21 45773189 45774294 TRPM2 chr22 24108000 24108140 CHCHD10 chr22 39106452 39106452 GTPBP1

chr3 3123308 3123308 IL5RA

chr3 12994821 12994919 IQSEC1

chr3 46394550 46395356 CCR2

chr3 49127364 49127364 QRICH1

chr3 58200471 58200569 DNASE1L3

chr3 71295684 71295684 FOXP1

chr3 127473936 127473936 MGLL

chr3 128998952 128998952 C3orf37

chr3 148581801 148581837 CPA3

chr3 156537970 156537970 LOC730091

185 chr3 176919597 176919597 TBL1XR1 chr3 184297380 184297522 EPHB3 chr3 193405999 193405999 OPA1 chr3 195603697 195603697 TNK2 chr3 196351986 196352142 LRRC33 chr4 1195845 1196179 SPON2 chr4 1294783 1295078 MAEA chr4 2431872 2432313 LOC402160 chr4 3374767 3374909 RGS12 chr4 24590014 24590014 DHX15 chr4 38859706 38859770 TLR6 chr4 87938832 87938832 AFF1 chr4 144266241 144266241 GAB1 chr4 157897045 157897045 PDGFC chr5 1089722 1089800 SLC12A7 chr5 1555791 1555887 SDHAP3 chr5 32297979 32297979 MTMR12 chr5 55148493 55148534 IL31RA chr5 67586170 67586258 PIK3R1 chr5 125694354 125694418 GRAMD3 chr5 137224213 137224284 MYOT chr5 143191565 143192067 HMHB1 chr5 153753045 153753045 GALNT10 chr5 169617918 169617918 C5orf58 chr5 177913434 177913485 COL23A1 chr6 6588693 6589075 LY86 chr6 16933816 16933816 FLJ23152 chr6 26030329 26030329 HIST1H3B chr6 28885444 28885568 TRIM27

186 chr6 30956397 30956440 MUC21 chr6 31088343 31088434 PSORS1C1 chr6 31629088 31629199 GPANK1 chr6 31654389 31654533 ABHD16A chr6 32099450 32099564 FKBPL chr6 32121055 32121393 PPT2 chr6 32135715 32136052 EGFL8 chr6 32810742 32810833 PSMB8 chr6 32825040 32825897 PSMB9 chr6 32905085 32905320 HLA-DMB chr6 41010111 41010316 UNC5CL chr6 41254433 41254471 TREM1 chr6 109812795 109812795 AKD1 chr6 135517041 135517046 MYB chr6 140168822 140168822 LOC100132735 chr6 142721804 142721804 GPR126 chr6 157876915 157877120 ZDHHC14 chr7 1545385 1545819 INTS1 chr7 2653651 2654120 IQCE chr7 30108301 30108301 PLEKHA8 chr7 33765409 33765409 BBS9 chr7 65419185 65419288 VKORC1L1 chr7 73645723 73645783 RFC2 chr7 80267619 80267943 CD36 chr7 101361395 101361745 MYL10 chr7 138347826 138348384 SVOPL chr7 138816336 138816336 TTC26 chr7 142659349 142659425 KEL chr8 1708438 1708526 CLN8

187 chr8 1870722 1870798 ARHGEF10 chr8 12608579 12608579 LONRF1 chr8 13105155 13105155 DLC1 chr8 33421410 33421410 RNF122 chr8 41654331 41655078 ANK1 chr8 42125496 42125496 IKBKB chr8 42623730 42623946 CHRNA6 chr8 68022262 68022262 CSPP1 chr8 99105576 99105576 C8orf47 chr8 128972450 128972829 PVT1 chr8 131368433 131368433 ASAP1 chr8 141312892 141312979 TRAPPC9 chr8 145643083 145643626 SLC39A4 chr9 132145105 132145105 C9orf106 chr9 136726359 136726575 VAV2 chr9 137277819 137277819 RXRA

188 Table 4.2. FAB type distribution for L-MPP-like and GMP-like AML samples

M0 M1 M2 M3 M4 M5 M6 M7 NA L-MPP 11 9 7 0 0 0 1 0 0 GMP 6 29 30 18 41 21 0 1 1

189 References

Abdel-Wahab, O., Adli, M., LaFave, L.M., Gao, J., Hricik, T., Shih, A.H., Pandey, S., Patel, J.P., Chung, Y.R., Koche, R., et al. (2012). ASXL1 mutations promote myeloid transformation through loss of PRC2-mediated gene repression. Cancer cell 22, 180-193.

Alharbi, R.A., Pettengell, R., Pandha, H.S., and Morgan, R. (2013). The role of HOX genes in normal hematopoiesis and acute leukemia. Leukemia 27, 1000-1008.

Aryee, M.J., Jaffe, A.E., Corrada-Bravo, H., Ladd-Acosta, C., Feinberg, A.P., Hansen, K.D., and Irizarry, R.A. (2014). Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics 30, 1363-1369.

Bach, C., Buhl, S., Mueller, D., Garcia-Cuellar, M.P., Maethner, E., and Slany, R.K. (2010). Leukemogenic transformation by HOXA cluster genes. Blood 115, 2910-2918.

Bartholdy, B., Christopeit, M., Will, B., Mo, Y., Barreyro, L., Yu, Y., Bhagat, T.D., Okoye-Okafor, U.C., Todorova, T.I., Greally, J.M., et al. (2014). HSC commitment- associated epigenetic signature is prognostic in acute myeloid leukemia. The Journal of clinical investigation 124, 1158-1167.

Bartke, T., Vermeulen, M., Xhemalce, B., Robson, S.C., Mann, M., and Kouzarides, T. (2010). Nucleosome-interacting proteins regulated by DNA and histone methylation. Cell 143, 470-484.

Bassil, C.F., Huang, Z., and Murphy, S.K. (2013). Bisulfite pyrosequencing. Methods in molecular biology 1049, 95-107.

Becker, A.J., Mc, C.E., and Till, J.E. (1963). Cytological demonstration of the clonal nature of spleen colonies derived from transplanted mouse marrow cells. Nature 197, 452-454.

Benetatos, L., Hatzimichael, E., Londin, E., Vartholomatos, G., Loher, P., Rigoutsos, I., and Briasoulis, E. (2013). The microRNAs within the DLK1-DIO3 genomic region: involvement in disease pathogenesis. Cellular and molecular life sciences : CMLS 70, 795-814.

190 Bennett, J.M., Catovsky, D., Daniel, M.T., Flandrin, G., Galton, D.A., Gralnick, H.R., and Sultan, C. (1976). Proposals for the classification of the acute leukaemias. French- American-British (FAB) co-operative group. British journal of haematology 33, 451-458.

Bennett, J.M., Catovsky, D., Daniel, M.T., Flandrin, G., Galton, D.A., Gralnick, H.R., and Sultan, C. (1985). Proposed revised criteria for the classification of acute myeloid leukemia. A report of the French-American-British Cooperative Group. Annals of internal medicine 103, 620-625.

Bernstein, B.E., Stamatoyannopoulos, J.A., Costello, J.F., Ren, B., Milosavljevic, A., Meissner, A., Kellis, M., Marra, M.A., Beaudet, A.L., Ecker, J.R., et al. (2010). The NIH Roadmap Epigenomics Mapping Consortium. Nature biotechnology 28, 1045-1048.

Bhutani, N., Brady, J.J., Damian, M., Sacco, A., Corbel, S.Y., and Blau, H.M. (2010). Reprogramming towards pluripotency requires AID-dependent DNA demethylation. Nature 463, 1042-1047.

Bird, A.P. (1986). CpG-rich islands and the function of DNA methylation. Nature 321, 209-213.

Bock, C., Beerman, I., Lien, W.H., Smith, Z.D., Gu, H., Boyle, P., Gnirke, A., Fuchs, E., Rossi, D.J., and Meissner, A. (2012). DNA methylation dynamics during in vivo differentiation of blood and skin stem cells. Molecular cell 47, 633-647.

Bonnet, D., and Dick, J.E. (1997). Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nature medicine 3, 730-737.

Brandeis, M., Frank, D., Keshet, I., Siegfried, Z., Mendelsohn, M., Nemes, A., Temper, V., Razin, A., and Cedar, H. (1994). Sp1 elements protect a CpG island from de novo methylation. Nature 371, 435-438.

Brinkman, A.B., Gu, H., Bartels, S.J., Zhang, Y., Matarese, F., Simmer, F., Marks, H., Bock, C., Gnirke, A., Meissner, A., et al. (2012). Sequential ChIP-bisulfite sequencing enables direct genome-scale investigation of chromatin and DNA methylation cross-talk. Genome research 22, 1128-1138.

Broske, A.M., Vockentanz, L., Kharazi, S., Huska, M.R., Mancini, E., Scheller, M., Kuhl, C., Enns, A., Prinz, M., Jaenisch, R., et al. (2009). DNA methylation protects hematopoietic stem cell multipotency from myeloerythroid restriction. Nature genetics 41, 1207-1215.

191 Bullinger, L., Ehrich, M., Dohner, K., Schlenk, R.F., Dohner, H., Nelson, M.R., and van den Boom, D. (2010). Quantitative DNA methylation predicts survival in adult acute myeloid leukemia. Blood 115, 636-642.

Busque, L., Patel, J.P., Figueroa, M.E., Vasanthakumar, A., Provost, S., Hamilou, Z., Mollica, L., Li, J., Viale, A., Heguy, A., et al. (2012). Recurrent somatic TET2 mutations in normal elderly individuals with clonal hematopoiesis. Nature genetics 44, 1179-1181.

Byrd, J.C., Mrozek, K., Dodge, R.K., Carroll, A.J., Edwards, C.G., Arthur, D.C., Pettenati, M.J., Patil, S.R., Rao, K.W., Watson, M.S., et al. (2002). Pretreatment cytogenetic abnormalities are predictive of induction success, cumulative incidence of relapse, and overall survival in adult patients with de novo acute myeloid leukemia: results from Cancer and Leukemia Group B (CALGB 8461). Blood 100, 4325-4336.

Cancer Genome Atlas Research, N. (2013). Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. The New England journal of medicine 368, 2059-2074.

Challen, G.A., Sun, D., Jeong, M., Luo, M., Jelinek, J., Berg, J.S., Bock, C., Vasanthakumar, A., Gu, H., Xi, Y., et al. (2012). Dnmt3a is essential for hematopoietic stem cell differentiation. Nature genetics 44, 23-31.

Chao, M.P., Seita, J., and Weissman, I.L. (2008). Establishment of a normal hematopoietic and leukemia stem cell hierarchy. Cold Spring Harbor symposia on quantitative biology 73, 439-449.

Chung, K.Y., Morrone, G., Schuringa, J.J., Plasilova, M., Shieh, J.H., Zhang, Y., Zhou, P., and Moore, M.A. (2006). Enforced expression of NUP98-HOXA9 in human CD34(+) cells enhances stem cell proliferation. Cancer Res 66, 11781-11791.

Civin, C.I., Strauss, L.C., Brovall, C., Fackler, M.J., Schwartz, J.F., and Shaper, J.H. (1984). Antigenic analysis of hematopoiesis. III. A hematopoietic progenitor cell surface antigen defined by a monoclonal antibody raised against KG-1a cells. Journal of immunology 133, 157-165.

Clark, S.J., Harrison, J., Paul, C.L., and Frommer, M. (1994). High sensitivity mapping of methylated cytosines. Nucleic acids research 22, 2990-2997.

Conerly, M.L., Teves, S.S., Diolaiti, D., Ulrich, M., Eisenman, R.N., and Henikoff, S. (2010). Changes in H2A.Z occupancy and DNA methylation during B-cell lymphomagenesis. Genome research 20, 1383-1390.

192 Corces-Zimmerman, M.R., Hong, W.J., Weissman, I.L., Medeiros, B.C., and Majeti, R. (2014). Preleukemic mutations in human acute myeloid leukemia affect epigenetic regulators and persist in remission. Proceedings of the National Academy of Sciences of the United States of America 111, 2548-2553.

Cortazar, D., Kunz, C., Saito, Y., Steinacher, R., and Schar, P. (2007). The enigmatic thymine DNA glycosylase. DNA repair 6, 489-504.

Cortellino, S., Xu, J., Sannai, M., Moore, R., Caretti, E., Cigliano, A., Le Coz, M., Devarajan, K., Wessels, A., Soprano, D., et al. (2011). Thymine DNA glycosylase is essential for active DNA demethylation by linked deamination-base excision repair. Cell 146, 67-79.

Daigle, S.R., Olhava, E.J., Therkelsen, C.A., Basavapathruni, A., Jin, L., Boriack-Sjodin, P.A., Allain, C.J., Klaus, C.R., Raimondi, A., Scott, M.P., et al. (2013). Potent inhibition of DOT1L as treatment of MLL-fusion leukemia. Blood 122, 1017-1025.

Daigle, S.R., Olhava, E.J., Therkelsen, C.A., Majer, C.R., Sneeringer, C.J., Song, J., Johnston, L.D., Scott, M.P., Smith, J.J., Xiao, Y., et al. (2011). Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor. Cancer cell 20, 53- 65.

Dang, L., White, D.W., Gross, S., Bennett, B.D., Bittinger, M.A., Driggers, E.M., Fantin, V.R., Jang, H.G., Jin, S., Keenan, M.C., et al. (2009). Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462, 739-744. de Jonge, H.J., Woolthuis, C.M., Vos, A.Z., Mulder, A., van den Berg, E., Kluin, P.M., van der Weide, K., de Bont, E.S., Huls, G., Vellenga, E., et al. (2011). Gene expression profiling in the leukemic stem cell-enriched CD34+ fraction identifies target genes that predict prognosis in normal karyotype AML. Leukemia 25, 1825-1833.

Deaton, A.M., and Bird, A. (2011). CpG islands and the regulation of transcription. Genes & development 25, 1010-1022.

Dedeurwaerder, S., Defrance, M., Calonne, E., Denis, H., Sotiriou, C., and Fuks, F. (2011). Evaluation of the Infinium Methylation 450K technology. Epigenomics 3, 771- 784.

Deneberg, S., Guardiola, P., Lennartsson, A., Qu, Y., Gaidzik, V., Blanchet, O., Karimi, M., Bengtzen, S., Nahi, H., Uggla, B., et al. (2011). Prognostic DNA methylation

193 patterns in cytogenetically normal acute myeloid leukemia are predefined by stem cell chromatin marks. Blood 118, 5573-5582.

DiGiusto, D., Chen, S., Combs, J., Webb, S., Namikawa, R., Tsukamoto, A., Chen, B.P., and Galy, A.H. (1994). Human fetal bone marrow early progenitors for T, B, and myeloid cells are found exclusively in the population expressing high levels of CD34. Blood 84, 421-432.

Dohner, H. (2007). Implication of the molecular characterization of acute myeloid leukemia. Hematology / the Education Program of the American Society of Hematology American Society of Hematology Education Program, 412-419.

Doi, A., Park, I.H., Wen, B., Murakami, P., Aryee, M.J., Irizarry, R., Herb, B., Ladd- Acosta, C., Rho, J., Loewer, S., et al. (2009). Differential methylation of tissue- and cancer-specific CpG island shores distinguishes human induced pluripotent stem cells, embryonic stem cells and fibroblasts. Nature genetics 41, 1350-1353.

Doulatov, S., Notta, F., Eppert, K., Nguyen, L.T., Ohashi, P.S., and Dick, J.E. (2010). Revised map of the human progenitor hierarchy shows the origin of macrophages and dendritic cells in early lymphoid development. Nature immunology 11, 585-593.

Doulatov, S., Notta, F., Laurenti, E., and Dick, J.E. (2012). Hematopoiesis: a human perspective. Cell stem cell 10, 120-136.

Drabkin, H.A., Parsy, C., Ferguson, K., Guilhot, F., Lacotte, L., Roy, L., Zeng, C., Baron, A., Hunger, S.P., Varella-Garcia, M., et al. (2002). Quantitative HOX expression in chromosomally defined subsets of acute myelogenous leukemia. Leukemia 16, 186-195.

Dufour, A., Schneider, F., Metzeler, K.H., Hoster, E., Schneider, S., Zellmeier, E., Benthaus, T., Sauerland, M.C., Berdel, W.E., Buchner, T., et al. (2010). Acute myeloid leukemia with biallelic CEBPA gene mutations and normal karyotype represents a distinct genetic entity associated with a favorable clinical outcome. J Clin Oncol 28, 570- 577.

Eppert, K., Takenaka, K., Lechman, E.R., Waldron, L., Nilsson, B., van Galen, P., Metzeler, K.H., Poeppl, A., Ling, V., Beyene, J., et al. (2011). Stem cell gene expression programs influence clinical outcome in human leukemia. Nature medicine 17, 1086-1093.

194 Ernst, T., Chase, A.J., Score, J., Hidalgo-Curtis, C.E., Bryant, C., Jones, A.V., Waghorn, K., Zoi, K., Ross, F.M., Reiter, A., et al. (2010). Inactivating mutations of the histone methyltransferase gene EZH2 in myeloid disorders. Nature genetics 42, 722-726.

Estey, E.H. (2013). Epigenetics in clinical practice: the examples of azacitidine and decitabine in myelodysplasia and acute myeloid leukemia. Leukemia 27, 1803-1812.

Feinberg, A.P., Ohlsson, R., and Henikoff, S. (2006). The epigenetic progenitor origin of human cancer. Nature reviews Genetics 7, 21-33.

Figueroa, M.E., Abdel-Wahab, O., Lu, C., Ward, P.S., Patel, J., Shih, A., Li, Y., Bhagwat, N., Vasanthakumar, A., Fernandez, H.F., et al. (2010a). Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer cell 18, 553-567.

Figueroa, M.E., Lugthart, S., Li, Y., Erpelinck-Verschueren, C., Deng, X., Christos, P.J., Schifano, E., Booth, J., van Putten, W., Skrabanek, L., et al. (2010b). DNA methylation signatures identify biologically distinct subtypes in acute myeloid leukemia. Cancer cell 17, 13-27.

Frommer, M., McDonald, L.E., Millar, D.S., Collis, C.M., Watt, F., Grigg, G.W., Molloy, P.L., and Paul, C.L. (1992). A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proceedings of the National Academy of Sciences of the United States of America 89, 1827-1831.

Genovese, G., Kahler, A.K., Handsaker, R.E., Lindberg, J., Rose, S.A., Bakhoum, S.F., Chambert, K., Mick, E., Neale, B.M., Fromer, M., et al. (2014). Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. The New England journal of medicine 371, 2477-2487.

Gentles, A.J., Plevritis, S.K., Majeti, R., and Alizadeh, A.A. (2010). Association of a leukemic stem cell gene expression signature with clinical outcomes in acute myeloid leukemia. Jama 304, 2706-2715.

Goardon, N., Marchi, E., Atzberger, A., Quek, L., Schuh, A., Soneji, S., Woll, P., Mead, A., Alford, K.A., Rout, R., et al. (2011). Coexistence of LMPP-like and GMP-like leukemia stem cells in acute myeloid leukemia. Cancer cell 19, 138-152.

Goldberg, A.D., Allis, C.D., and Bernstein, E. (2007). Epigenetics: a landscape takes shape. Cell 128, 635-638.

195 Golub, T.R., Slonim, D.K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J.P., Coller, H., Loh, M.L., Downing, J.R., Caligiuri, M.A., et al. (1999). Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286, 531-537.

Grimwade, D., Walker, H., Oliver, F., Wheatley, K., Harrison, C., Harrison, G., Rees, J., Hann, I., Stevens, R., Burnett, A., et al. (1998). The importance of diagnostic cytogenetics on outcome in AML: analysis of 1,612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children's Leukaemia Working Parties. Blood 92, 2322-2333.

Hansen, K.D., Langmead, B., and Irizarry, R.A. (2012). BSmooth: from whole genome bisulfite sequencing reads to differentially methylated regions. Genome biology 13, R83.

Hansen, K.D., Timp, W., Bravo, H.C., Sabunciyan, S., Langmead, B., McDonald, O.G., Wen, B., Wu, H., Liu, Y., Diep, D., et al. (2011). Increased methylation variation in epigenetic domains across cancer types. Nature genetics 43, 768-775.

He, Y.F., Li, B.Z., Li, Z., Liu, P., Wang, Y., Tang, Q., Ding, J., Jia, Y., Chen, Z., Li, L., et al. (2011). Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian DNA. Science 333, 1303-1307.

Herb, B.R., Wolschin, F., Hansen, K.D., Aryee, M.J., Langmead, B., Irizarry, R., Amdam, G.V., and Feinberg, A.P. (2012). Reversible switching between epigenetic states in honeybee behavioral subcastes. Nature neuroscience 15, 1371-1373.

Heuser, M., Yun, H., Berg, T., Yung, E., Argiropoulos, B., Kuchenbauer, F., Park, G., Hamwi, I., Palmqvist, L., Lai, C.K., et al. (2011). Cell of origin in AML: susceptibility to MN1-induced transformation is regulated by the MEIS1/AbdB-like HOX protein complex. Cancer cell 20, 39-52.

Hnisz, D., Abraham, B.J., Lee, T.I., Lau, A., Saint-Andre, V., Sigova, A.A., Hoke, H.A., and Young, R.A. (2013). Super-enhancers in the control of cell identity and disease. Cell 155, 934-947.

Hosen, N., Park, C.Y., Tatsumi, N., Oji, Y., Sugiyama, H., Gramatzki, M., Krensky, A.M., and Weissman, I.L. (2007). CD96 is a leukemic stem cell-specific marker in human acute myeloid leukemia. Proceedings of the National Academy of Sciences of the United States of America 104, 11008-11013.

196 Huang, Y., Pastor, W.A., Shen, Y., Tahiliani, M., Liu, D.R., and Rao, A. (2010). The behaviour of 5-hydroxymethylcytosine in bisulfite sequencing. PloS one 5, e8888.

Irizarry, R.A., Ladd-Acosta, C., Carvalho, B., Wu, H., Brandenburg, S.A., Jeddeloh, J.A., Wen, B., and Feinberg, A.P. (2008). Comprehensive high-throughput arrays for relative methylation (CHARM). Genome research 18, 780-790.

Irizarry, R.A., Ladd-Acosta, C., Wen, B., Wu, Z., Montano, C., Onyango, P., Cui, H., Gabo, K., Rongione, M., Webster, M., et al. (2009). The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nature genetics 41, 178-186.

Ito, S., D'Alessio, A.C., Taranova, O.V., Hong, K., Sowers, L.C., and Zhang, Y. (2010). Role of Tet proteins in 5mC to 5hmC conversion, ES-cell self-renewal and inner cell mass specification. Nature 466, 1129-1133.

Ito, S., Shen, L., Dai, Q., Wu, S.C., Collins, L.B., Swenberg, J.A., He, C., and Zhang, Y. (2011). Tet proteins can convert 5-methylcytosine to 5-formylcytosine and 5- carboxylcytosine. Science 333, 1300-1303.

Jaffe, A.E., Murakami, P., Lee, H., Leek, J.T., Fallin, M.D., Feinberg, A.P., and Irizarry, R.A. (2012). Bump hunting to identify differentially methylated regions in epigenetic epidemiology studies. Int J Epidemiol 41, 200-209.

Jaiswal, S., Fontanillas, P., Flannick, J., Manning, A., Grauman, P.V., Mar, B.G., Lindsley, R.C., Mermel, C.H., Burtt, N., Chavez, A., et al. (2014). Age-related clonal hematopoiesis associated with adverse outcomes. The New England journal of medicine 371, 2488-2498.

Jamieson, C.H., Ailles, L.E., Dylla, S.J., Muijtjens, M., Jones, C., Zehnder, J.L., Gotlib, J., Li, K., Manz, M.G., Keating, A., et al. (2004). Granulocyte-macrophage progenitors as candidate leukemic stem cells in blast-crisis CML. The New England journal of medicine 351, 657-667.

Jan, M., Chao, M.P., Cha, A.C., Alizadeh, A.A., Gentles, A.J., Weissman, I.L., and Majeti, R. (2011). Prospective separation of normal and leukemic stem cells based on differential expression of TIM3, a human acute myeloid leukemia stem cell marker. Proceedings of the National Academy of Sciences of the United States of America 108, 5009-5014.

197 Jan, M., Snyder, T.M., Corces-Zimmerman, M.R., Vyas, P., Weissman, I.L., Quake, S.R., and Majeti, R. (2012). Clonal evolution of preleukemic hematopoietic stem cells precedes human acute myeloid leukemia. Science translational medicine 4, 149ra118.

Ji, H., Ehrlich, L.I., Seita, J., Murakami, P., Doi, A., Lindau, P., Lee, H., Aryee, M.J., Irizarry, R.A., Kim, K., et al. (2010). Comprehensive methylome map of lineage commitment from haematopoietic progenitors. Nature 467, 338-342.

Jones, P.A. (2012). Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nature reviews Genetics 13, 484-492.

Kikushige, Y., Shima, T., Takayanagi, S., Urata, S., Miyamoto, T., Iwasaki, H., Takenaka, K., Teshima, T., Tanaka, T., Inagaki, Y., et al. (2010). TIM-3 is a promising target to selectively kill acute myeloid leukemia stem cells. Cell stem cell 7, 708-717.

Kim, K., Doi, A., Wen, B., Ng, K., Zhao, R., Cahan, P., Kim, J., Aryee, M.J., Ji, H., Ehrlich, L.I., et al. (2010). Epigenetic memory in induced pluripotent stem cells. Nature 467, 285-290.

Kim, K., Zhao, R., Doi, A., Ng, K., Unternaehrer, J., Cahan, P., Huo, H., Loh, Y.H., Aryee, M.J., Lensch, M.W., et al. (2011). Donor cell type can influence the epigenome and differentiation potential of human induced pluripotent stem cells. Nature biotechnology 29, 1117-1119.

Klebanoff, S.J. (2005). Myeloperoxidase: friend and foe. Journal of leukocyte biology 77, 598-625.

Ko, M., Huang, Y., Jankowska, A.M., Pape, U.J., Tahiliani, M., Bandukwala, H.S., An, J., Lamperti, E.D., Koh, K.P., Ganetzky, R., et al. (2010). Impaired hydroxylation of 5- methylcytosine in myeloid cancers with mutant TET2. Nature 468, 839-843.

Kohli, R.M., and Zhang, Y. (2013). TET enzymes, TDG and the dynamics of DNA demethylation. Nature 502, 472-479.

Kozar, K., and Sicinski, P. (2005). Cell cycle progression without cyclin D-CDK4 and cyclin D-CDK6 complexes. Cell cycle 4, 388-391.

Krause, D.S., Fackler, M.J., Civin, C.I., and May, W.S. (1996). CD34: structure, biology, and clinical utility. Blood 87, 1-13.

198 Kreso, A., and Dick, J.E. (2014). Evolution of the cancer stem cell model. Cell stem cell 14, 275-291.

Kriaucionis, S., and Heintz, N. (2009). The nuclear DNA base 5-hydroxymethylcytosine is present in Purkinje neurons and the brain. Science 324, 929-930.

Krivtsov, A.V., Figueroa, M.E., Sinha, A.U., Stubbs, M.C., Feng, Z., Valk, P.J., Delwel, R., Dohner, K., Bullinger, L., Kung, A.L., et al. (2013). Cell of origin determines clinically relevant subtypes of MLL-rearranged AML. Leukemia 27, 852-860.

Kulis, M., Heath, S., Bibikova, M., Queiros, A.C., Navarro, A., Clot, G., Martinez-Trillos, A., Castellano, G., Brun-Heath, I., Pinyol, M., et al. (2012). Epigenomic analysis detects widespread gene-body DNA hypomethylation in chronic lymphocytic leukemia. Nature genetics 44, 1236-1242.

Kumar, R., DiMenna, L., Schrode, N., Liu, T.C., Franck, P., Munoz-Descalzo, S., Hadjantonakis, A.K., Zarrin, A.A., Chaudhuri, J., Elemento, O., et al. (2013). AID stabilizes stem-cell phenotype by removing epigenetic memory of pluripotency genes. Nature 500, 89-92.

Laird, P.W. (2010). Principles and challenges of genomewide DNA methylation analysis. Nature reviews Genetics 11, 191-203.

Lapidot, T., Sirard, C., Vormoor, J., Murdoch, B., Hoang, T., Caceres-Cortes, J., Minden, M., Paterson, B., Caligiuri, M.A., and Dick, J.E. (1994). A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature 367, 645-648.

Laurent, L., Wong, E., Li, G., Huynh, T., Tsirigos, A., Ong, C.T., Low, H.M., Kin Sung, K.W., Rigoutsos, I., Loring, J., et al. (2010). Dynamic changes in the human methylome during differentiation. Genome research 20, 320-331.

Lehnertz, B., Pabst, C., Su, L., Miller, M., Liu, F., Yi, L., Zhang, R., Krosl, J., Yung, E., Kirschner, J., et al. (2014). The methyltransferase G9a regulates HoxA9-dependent transcription in AML. Genes Dev 28, 317-327.

Leonhardt, H., Page, A.W., Weier, H.U., and Bestor, T.H. (1992). A targeting sequence directs DNA methyltransferase to sites of DNA replication in mammalian nuclei. Cell 71, 865-873.

199 Ley, T.J., Ding, L., Walter, M.J., McLellan, M.D., Lamprecht, T., Larson, D.E., Kandoth, C., Payton, J.E., Baty, J., Welch, J., et al. (2010). DNMT3A mutations in acute myeloid leukemia. The New England journal of medicine 363, 2424-2433.

Li, E., Bestor, T.H., and Jaenisch, R. (1992). Targeted mutation of the DNA methyltransferase gene results in embryonic lethality. Cell 69, 915-926.

Lister, R., Mukamel, E.A., Nery, J.R., Urich, M., Puddifoot, C.A., Johnson, N.D., Lucero, J., Huang, Y., Dwork, A.J., Schultz, M.D., et al. (2013). Global epigenomic reconfiguration during mammalian brain development. Science 341, 1237905.

Lister, R., Pelizzola, M., Dowen, R.H., Hawkins, R.D., Hon, G., Tonti-Filippini, J., Nery, J.R., Lee, L., Ye, Z., Ngo, Q.M., et al. (2009). Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462, 315-322.

Losada, A. (2014). Cohesin in cancer: chromosome segregation and beyond. Nat Rev Cancer 14, 389-393.

Lowenberg, B., Downing, J.R., and Burnett, A. (1999). Acute myeloid leukemia. The New England journal of medicine 341, 1051-1062.

Lund, K., Adams, P.D., and Copland, M. (2014). EZH2 in normal and malignant hematopoiesis. Leukemia 28, 44-49.

Macleod, D., Charlton, J., Mullins, J., and Bird, A.P. (1994). Sp1 sites in the mouse aprt gene promoter are required to prevent methylation of the CpG island. Genes & development 8, 2282-2292.

Maiti, A., and Drohat, A.C. (2011). Thymine DNA glycosylase can rapidly excise 5- formylcytosine and 5-carboxylcytosine: potential implications for active demethylation of CpG sites. The Journal of biological chemistry 286, 35334-35338.

Majeti, R., Becker, M.W., Tian, Q., Lee, T.L., Yan, X., Liu, R., Chiang, J.H., Hood, L., Clarke, M.F., and Weissman, I.L. (2009a). Dysregulated gene expression networks in human acute myelogenous leukemia stem cells. Proceedings of the National Academy of Sciences of the United States of America 106, 3396-3401.

Majeti, R., Chao, M.P., Alizadeh, A.A., Pang, W.W., Jaiswal, S., Gibbs, K.D., Jr., van Rooijen, N., and Weissman, I.L. (2009b). CD47 is an adverse prognostic factor and

200 therapeutic antibody target on human acute myeloid leukemia stem cells. Cell 138, 286- 299.

Majeti, R., Park, C.Y., and Weissman, I.L. (2007). Identification of a hierarchy of multipotent hematopoietic progenitors in human cord blood. Cell stem cell 1, 635-645.

Malumbres, M., Sotillo, R., Santamaria, D., Galan, J., Cerezo, A., Ortega, S., Dubus, P., and Barbacid, M. (2004). Mammalian cells cycle without the D-type cyclin-dependent kinases Cdk4 and Cdk6. Cell 118, 493-504.

Marcucci, G., Maharry, K., Wu, Y.Z., Radmacher, M.D., Mrozek, K., Margeson, D., Holland, K.B., Whitman, S.P., Becker, H., Schwind, S., et al. (2010). IDH1 and IDH2 gene mutations identify novel molecular subsets within de novo cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 28, 2348-2355.

Mardis, E.R., Ding, L., Dooling, D.J., Larson, D.E., McLellan, M.D., Chen, K., Koboldt, D.C., Fulton, R.S., Delehaunty, K.D., McGrath, S.D., et al. (2009). Recurring mutations found by sequencing an acute myeloid leukemia genome. The New England journal of medicine 361, 1058-1066.

Martelli, M.P., Pettirossi, V., Thiede, C., Bonifacio, E., Mezzasoma, F., Cecchini, D., Pacini, R., Tabarrini, A., Ciurnelli, R., Gionfriddo, I., et al. (2010). CD34+ cells from AML with mutated NPM1 harbor cytoplasmic mutated nucleophosmin and generate leukemia in immunocompromised mice. Blood 116, 3907-3922.

Mestas, J., and Hughes, C.C. (2004). Of mice and not men: differences between mouse and human immunology. Journal of immunology 172, 2731-2738.

Metzeler, K.H., Hummel, M., Bloomfield, C.D., Spiekermann, K., Braess, J., Sauerland, M.C., Heinecke, A., Radmacher, M., Marcucci, G., Whitman, S.P., et al. (2008). An 86- probe-set gene-expression signature predicts survival in cytogenetically normal acute myeloid leukemia. Blood 112, 4193-4201.

Milne, T.A., Briggs, S.D., Brock, H.W., Martin, M.E., Gibbs, D., Allis, C.D., and Hess, J.L. (2002). MLL targets SET domain methyltransferase activity to Hox gene promoters. Mol Cell 10, 1107-1117.

Moran-Crusio, K., Reavie, L., Shih, A., Abdel-Wahab, O., Ndiaye-Lobry, D., Lobry, C., Figueroa, M.E., Vasanthakumar, A., Patel, J., Zhao, X., et al. (2011). Tet2 loss leads to

201 increased hematopoietic stem cell self-renewal and myeloid transformation. Cancer cell 20, 11-24.

Nakamura, T., Largaespada, D.A., Lee, M.P., Johnson, L.A., Ohyashiki, K., Toyama, K., Chen, S.J., Willman, C.L., Chen, I.M., Feinberg, A.P., et al. (1996). Fusion of the nucleoporin gene NUP98 to HOXA9 by the chromosome translocation t(7;11)(p15;p15) in human myeloid leukaemia. Nat Genet 12, 154-158.

Okano, M., Bell, D.W., Haber, D.A., and Li, E. (1999). DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 99, 247-257.

Otani, J., Nankumo, T., Arita, K., Inamoto, S., Ariyoshi, M., and Shirakawa, M. (2009). Structural basis for recognition of H3K4 methylation status by the DNA methyltransferase 3A ATRX-DNMT3-DNMT3L domain. EMBO reports 10, 1235-1241.

Patel, J.P., Gonen, M., Figueroa, M.E., Fernandez, H., Sun, Z., Racevskis, J., Van Vlierberghe, P., Dolgalev, I., Thomas, S., Aminova, O., et al. (2012). Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. The New England journal of medicine 366, 1079-1089.

Polak, P., Karlic, R., Koren, A., Thurman, R., Sandstrom, R., Lawrence, M.S., Reynolds, A., Rynes, E., Vlahovicek, K., Stamatoyannopoulos, J.A., et al. (2015). Cell-of-origin chromatin organization shapes the mutational landscape of cancer. Nature 518, 360-364.

Popp, C., Dean, W., Feng, S., Cokus, S.J., Andrews, S., Pellegrini, M., Jacobsen, S.E., and Reik, W. (2010). Genome-wide erasure of DNA methylation in mouse primordial germ cells is affected by AID deficiency. Nature 463, 1101-1105.

Quivoron, C., Couronne, L., Della Valle, V., Lopez, C.K., Plo, I., Wagner-Ballon, O., Do Cruzeiro, M., Delhommeau, F., Arnulf, B., Stern, M.H., et al. (2011). TET2 inactivation results in pleiotropic hematopoietic abnormalities in mouse and is a recurrent event during human lymphomagenesis. Cancer cell 20, 25-38.

Ramsahoye, B.H., Biniszkiewicz, D., Lyko, F., Clark, V., Bird, A.P., and Jaenisch, R. (2000). Non-CpG methylation is prevalent in embryonic stem cells and may be mediated by DNA methyltransferase 3a. Proceedings of the National Academy of Sciences of the United States of America 97, 5237-5242.

202 Roadmap Epigenomics, C., Kundaje, A., Meuleman, W., Ernst, J., Bilenky, M., Yen, A., Heravi-Moussavi, A., Kheradpour, P., Zhang, Z., Wang, J., et al. (2015). Integrative analysis of 111 reference human epigenomes. Nature 518, 317-330.

Saito, Y., Kitamura, H., Hijikata, A., Tomizawa-Murasawa, M., Tanaka, S., Takagi, S., Uchida, N., Suzuki, N., Sone, A., Najima, Y., et al. (2010). Identification of therapeutic targets for quiescent, chemotherapy-resistant human leukemia stem cells. Science translational medicine 2, 17ra19.

Samudio, I., Harmancey, R., Fiegl, M., Kantarjian, H., Konopleva, M., Korchin, B., Kaluarachchi, K., Bornmann, W., Duvvuri, S., Taegtmeyer, H., et al. (2010). Pharmacologic inhibition of fatty acid oxidation sensitizes human leukemia cells to apoptosis induction. J Clin Invest 120, 142-156.

Sarry, J.E., Murphy, K., Perry, R., Sanchez, P.V., Secreto, A., Keefer, C., Swider, C.R., Strzelecki, A.C., Cavelier, C., Recher, C., et al. (2011). Human acute myelogenous leukemia stem cells are rare and heterogeneous when assayed in NOD/SCID/IL2Rgammac-deficient mice. J Clin Invest 121, 384-395.

Schultz, M.D., He, Y., Whitaker, J.W., Hariharan, M., Mukamel, E.A., Leung, D., Rajagopal, N., Nery, J.R., Urich, M.A., Chen, H., et al. (2015). Human body epigenome maps reveal noncanonical DNA methylation variation. Nature.

Shih, A.H., Abdel-Wahab, O., Patel, J.P., and Levine, R.L. (2012). The role of mutations in epigenetic regulators in myeloid malignancies. Nature reviews Cancer 12, 599-612.

Shlush, L.I., Zandi, S., Mitchell, A., Chen, W.C., Brandwein, J.M., Gupta, V., Kennedy, J.A., Schimmer, A.D., Schuh, A.C., Yee, K.W., et al. (2014). Identification of pre- leukaemic haematopoietic stem cells in acute leukaemia. Nature 506, 328-333.

Slovak, M.L., Kopecky, K.J., Cassileth, P.A., Harrington, D.H., Theil, K.S., Mohamed, A., Paietta, E., Willman, C.L., Head, D.R., Rowe, J.M., et al. (2000). Karyotypic analysis predicts outcome of preremission and postremission therapy in adult acute myeloid leukemia: a Southwest Oncology Group/Eastern Cooperative Oncology Group Study. Blood 96, 4075-4083.

Smith, Z.D., and Meissner, A. (2013). DNA methylation: roles in mammalian development. Nature reviews Genetics 14, 204-220.

203 Tadokoro, Y., Ema, H., Okano, M., Li, E., and Nakauchi, H. (2007). De novo DNA methyltransferase is essential for self-renewal, but not for differentiation, in hematopoietic stem cells. The Journal of experimental medicine 204, 715-722.

Tahiliani, M., Koh, K.P., Shen, Y., Pastor, W.A., Bandukwala, H., Brudno, Y., Agarwal, S., Iyer, L.M., Liu, D.R., Aravind, L., et al. (2009). Conversion of 5-methylcytosine to 5- hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science 324, 930- 935.

Takeda, A., Goolsby, C., and Yaseen, N.R. (2006). NUP98-HOXA9 induces long-term proliferation and blocks differentiation of primary human CD34+ hematopoietic cells. Cancer Res 66, 6628-6637.

Taussig, D.C., Miraki-Moud, F., Anjos-Afonso, F., Pearce, D.J., Allen, K., Ridler, C., Lillington, D., Oakervee, H., Cavenagh, J., Agrawal, S.G., et al. (2008). Anti-CD38 antibody-mediated clearance of human repopulating cells masks the heterogeneity of leukemia-initiating cells. Blood 112, 568-575.

Taussig, D.C., Vargaftig, J., Miraki-Moud, F., Griessinger, E., Sharrock, K., Luke, T., Lillington, D., Oakervee, H., Cavenagh, J., Agrawal, S.G., et al. (2010). Leukemia- initiating cells from some acute myeloid leukemia patients with mutated nucleophosmin reside in the CD34(-) fraction. Blood 115, 1976-1984.

Thol, F., Bollin, R., Gehlhaar, M., Walter, C., Dugas, M., Suchanek, K.J., Kirchner, A., Huang, L., Chaturvedi, A., Wichmann, M., et al. (2014). Mutations in the cohesin complex in acute myeloid leukemia: clinical and prognostic implications. Blood 123, 914-920.

Thorsteinsdottir, U., Mamo, A., Kroon, E., Jerome, L., Bijl, J., Lawrence, H.J., Humphries, K., and Sauvageau, G. (2002). Overexpression of the myeloid leukemia- associated Hoxa9 gene in bone marrow cells induces stem cell expansion. Blood 99, 121- 129.

Till, J.E., and Mc, C.E. (1961). A direct measurement of the radiation sensitivity of normal mouse bone marrow cells. Radiation research 14, 213-222.

Timp, W., Bravo, H.C., McDonald, O.G., Goggins, M., Umbricht, C., Zeiger, M., Feinberg, A.P., and Irizarry, R.A. (2014). Large hypomethylated blocks as a universal defining epigenetic alteration in human solid tumors. Genome medicine 6, 61.

204 Timp, W., and Feinberg, A.P. (2013). Cancer as a dysregulated epigenome allowing cellular growth advantage at the expense of the host. Nature reviews Cancer 13, 497-510.

Trowbridge, J.J., Snow, J.W., Kim, J., and Orkin, S.H. (2009). DNA methyltransferase 1 is essential for and uniquely regulates hematopoietic stem and progenitor cells. Cell stem cell 5, 442-449.

Valent, P., and Zuber, J. (2014). BRD4: a BET(ter) target for the treatment of AML? Cell cycle 13, 689-690.

Valk, P.J., Verhaak, R.G., Beijen, M.A., Erpelinck, C.A., Barjesteh van Waalwijk van Doorn-Khosrovani, S., Boer, J.M., Beverloo, H.B., Moorhouse, M.J., van der Spek, P.J., Lowenberg, B., et al. (2004). Prognostically useful gene-expression profiles in acute myeloid leukemia. N Engl J Med 350, 1617-1628. van Rhenen, A., van Dongen, G.A., Kelder, A., Rombouts, E.J., Feller, N., Moshaver, B., Stigter-van Walsum, M., Zweegman, S., Ossenkoppele, G.J., and Jan Schuurhuis, G. (2007). The novel AML stem cell associated antigen CLL-1 aids in discrimination between normal and leukemic stem cells. Blood 110, 2659-2666.

Vandiver, A.R., Irizarry, R.A., Hansen, K.D., Garza, L.A., Runarsson, A., Li, X., Chien, A.L., Wang, T.S., Leung, S.G., Kang, S., et al. (2015). Age and sun exposure-related widespread genomic blocks of hypomethylation in nonmalignant skin. Genome biology 16, 80.

Varley, K.E., Gertz, J., Bowling, K.M., Parker, S.L., Reddy, T.E., Pauli-Behn, F., Cross, M.K., Williams, B.A., Stamatoyannopoulos, J.A., Crawford, G.E., et al. (2013). Dynamic DNA methylation across diverse human cell lines and tissues. Genome research 23, 555- 567.

Vassiliou, G.S., Cooper, J.L., Rad, R., Li, J., Rice, S., Uren, A., Rad, L., Ellis, P., Andrews, R., Banerjee, R., et al. (2011). Mutant nucleophosmin and cooperating pathways drive leukemia initiation and progression in mice. Nat Genet 43, 470-475.

Waddington, C.H. (2012). The epigenotype. 1942. International journal of epidemiology 41, 10-13.

Ward, P.S., Patel, J., Wise, D.R., Abdel-Wahab, O., Bennett, B.D., Coller, H.A., Cross, J.R., Fantin, V.R., Hedvat, C.V., Perl, A.E., et al. (2010). The common feature of

205 leukemia-associated IDH1 and IDH2 mutations is a neomorphic enzyme activity converting alpha-ketoglutarate to 2-hydroxyglutarate. Cancer cell 17, 225-234.

Whyte, W.A., Orlando, D.A., Hnisz, D., Abraham, B.J., Lin, C.Y., Kagey, M.H., Rahl, P.B., Lee, T.I., and Young, R.A. (2013). Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell 153, 307-319.

Wilson, C.S., Davidson, G.S., Martin, S.B., Andries, E., Potter, J., Harvey, R., Ar, K., Xu, Y., Kopecky, K.J., Ankerst, D.P., et al. (2006). Gene expression profiling of adult acute myeloid leukemia identifies novel biologic clusters for risk classification and outcome prediction. Blood 108, 685-696.

Wouters, B.J., Lowenberg, B., Erpelinck-Verschueren, C.A., van Putten, W.L., Valk, P.J., and Delwel, R. (2009). Double CEBPA mutations, but not single CEBPA mutations, define a subgroup of acute myeloid leukemia with a distinctive gene expression profile that is uniquely associated with a favorable outcome. Blood 113, 3088-3091.

Xie, M., Lu, C., Wang, J., McLellan, M.D., Johnson, K.J., Wendl, M.C., McMichael, J.F., Schmidt, H.K., Yellapantula, V., Miller, C.A., et al. (2014). Age-related mutations associated with clonal hematopoietic expansion and malignancies. Nature medicine 20, 1472-1478.

Yan, X.J., Xu, J., Gu, Z.H., Pan, C.M., Lu, G., Shen, Y., Shi, J.Y., Zhu, Y.M., Tang, L., Zhang, X.W., et al. (2011). Exome sequencing identifies somatic mutations of DNA methyltransferase gene DNMT3A in acute monocytic leukemia. Nature genetics 43, 309- 315.

Yang, X., Han, H., De Carvalho, D.D., Lay, F.D., Jones, P.A., and Liang, G. (2014). Gene body methylation can alter gene expression and is a therapeutic target in cancer. Cancer cell 26, 577-590.

Yoder, J.A., Walsh, C.P., and Bestor, T.H. (1997). Cytosine methylation and the ecology of intragenomic parasites. Trends in genetics : TIG 13, 335-340.

Zuber, J., Shi, J., Wang, E., Rappaport, A.R., Herrmann, H., Sison, E.A., Magoon, D., Qi, J., Blatt, K., Wunderlich, M., et al. (2011). RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature 478, 524-528.

206 Curriculum Vitae

Namyoung Jung

Johns Hopkins School of Medicine 570 Rangos Building, Center for Epigenetics Phone: 443-683-2135 855 North Wolfe Street, Baltimore, MD 21205 E-mail: [email protected]

EDUCATION 2009-present Johns Hopkins University School of Medicine, Baltimore, MD PhD Candidate in Cellular and Molecular Medicine Graduate Program. Thesis advisor: Andrew P. Feinberg, MD.,MPH. 2003-2007 Pohang University of Science and Technology (POSTECH), Pohang, South Korea B.S. in Life Science, Summa Cum Laude. Thesis advisor: Yun-je Cho, PhD. 2005 University of California, Berkeley, Berkeley, CA Exchange student in Dept. of Molecular and Cellular Biology.

RESEARCH EXPERIENCE 2009-present Johns Hopkins University School of Medicine, Center for Epigenetics and Department of Medicine PhD candidate, Thesis advisor: Andrew P. Feinberg, MD.,MPH. Investigated the role of epigenetics, particularly DNA methylation in progenitor cell biology. Two major model systems were explored; Leukemia stem cells (LSCs) along with normal hematopoietic stem and progenitor cells (HSPCs), and induced pluripotent stem cells (iPSCs).

207 a. Epigenetic features of LSCs and HSPCs using Illumina Infinium HumanMethylation450 bead chip array. Collaboration with Ravindra Majeti, MD., PhD. at Stanford University School of Medicine. -Identified LSC epigenetic signature, DNA methylation differences that correlates with differential gene expression between LSCs and blasts. -Demonstrated prognostic power of LSC epigenetic signature for AML patients. -Discovered novel regulators for human hematopoiesis with genome-wide DNA methylation and gene expression analysis. -Identified subgroups of LSCs that reflect cell of origin with DNA methylation analysis. b. Epigenetic differences of iPSC clones derived by different non-integrating methods using mRNA, sendai virus and episome with whole genome bisulfite sequencing (WGBS) and chromatin immunoprecipitation sequencing (ChIP-seq) analysis. Collaboration with George Q. Daley, MD., PhD. at Harvard University School of Medicine. -Identified no significant DNA methylation differences among the iPSCs derived from different methods

2008-2009 Yonsei University, Center for Genome Regulation (Seoul, South Korea). Research assistant, Supervisor: Youngjoon Kim, PhD. -Optimized native ChIP and identified specific methylation patterns in obesity and cancer.

2007 Swiss Institute for Experimental Cancer Research (ISREC), Telomerase and Chromosome End Replication Laboratory (Lausanne, Switzerland). Summer research program student. Research advisor: Joachim Lingner, PhD. -Identified the function of Rat1, an exonuclease involved in DNA transcription termination, replication and telomere senescence in Saccharomyces. Pombe.

208 2006-2007 Pohang University of Science and Technology (POSTECH) (Pohang, South Korea) Undergraduate student, Thesis advisor: Yunje Cho, PhD. -Performed structural study of MCM10, a member of DNA replication complex in eukaryote.

2006 Pohang University of Science and Technology (POSTECH) (Pohang, South Korea) Undergraduate student, Research advisor: Sungho Ryu, PhD. -Studied the relationship between glycolysis and mTOR pathway.

2005 Pohang University of Science and Technology (POSTECH) (Pohang, South Korea) Undergraduate student, Research advisor: Byung-ha Oh, PhD. -Conducted preliminary structural study of autophagy complex.

HONORS & AWARDS 2014 Mogam Scholarship 2009-2014 Samsung Scholarship 2007 Academic Scholarship for EPFL (Ecole Polytechnique Federale de Lausanne) Summer Research Program 2007 Summa Cum Laude, POSTECH 2007 Nuri Scholarship for excellent research project, KRF (Korea Research Foundation) 2003-2007 Full Scholarship for talented undergraduate students in science major, KOSEF (Korea Science and Engineering Foundation)

TEACHING EXPERIENCE 2009-present Johns Hopkins University School of Medicine, Mentor for rotation students.

209 -Taught how to perform experiments such as bisulfite pyrosequencing, WGBS, comprehensive high-throughput arrays for relative methylation (CHARM), genomic DNA extraction. -Guided rotation students to make an academic poster and give a talk.

TALKS

Jung N*, Dai B*, Murakami P, Gentles AJ, Majeti R†, Feinberg AP†. Epigenetic signature of leukemia stem cells defines subgroups associated with clinical outcome in AML. The Centers of Excellence in Genomic Science (CEGS) annual meeting, 2014

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Comprehensive methylome map of human hematopoietic stem and progenitor cells. NHLBI Progenitor Cell Biology Consortium (PCBC) annual meeting, 2014

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Comprehensive methylome map of human hematopoietic stem and progenitor cells. NHLBI Progenitor Cell Biology Consortium (PCBC) annual meeting, 2013

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Comprehensive methylome map of human hematopoietic stem and progenitor cells. NHLBI Progenitor Cell Biology Consortium (PCBC) annual meeting, 2012

POSTER PRESENTATIONS

Jung N*, Dai B*, Gentles AJ, Murakami P, Majeti R†, Feinberg AP†. Epigenetic signature of leukemia stem cells defines subgroups associated with clinical outcome and cell of origin in AML. The American Society of Hematology annual meeting, 2014

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Comprehensive methylome map of human hematopoietic stem and progenitor cells. NHLBI Progenitor Cell Biology Consortium (PCBC) annual meeting, 2014

210

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Increased genome- wide epigenetic variation distinguishes acute myeloid leukemia from human normal hematopoiesis, The Centers of Excellence in Genomic Science (CEGS) annual meeting, 2013

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Comprehensive methylome map of human hematopoietic stem and progenitor cells. NHLBI Progenitor Cell Biology Consortium (PCBC) annual meeting, 2013

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Increased genome- wide epigenetic variation distinguishes acute myeloid leukemia from human normal hematopoiesis, The Centers of Excellence in Genomic Science (CEGS) annual meeting, 2012

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Increased genome- wide epigenetic variation distinguishes acute myeloid leukemia from human normal hematopoiesis, Wellcome Trust Epigenomics of Common Diseases Conference, 2012

Jung N*, Dai B*, Murakami P, Irizarry R, Majeti R†, Feinberg AP†. Comprehensive methylome map of human hematopoietic stem and progenitor cells. NHLBI Progenitor Cell Biology Consortium (PCBC) annual meeting, 2012

Jung N*, Dai B*, Ji H, Weissman I, Majeti R†, Feinberg AP†. Epigenetic change during acute myeloid leukemia progression. The Centers of Excellence in Genomic Science (CEGS) annual meeting, 2011

PUBLICATIONS

Jung N*, Li X*, Honaker C, Siegel PB, Andersson L, Feinberg AP. Widespread mutational change driving DNA methylation in size selection of domestic chickens. (In preparation)

211 Jung N*, Dai B*, Gentles AJ, Majeti R†, Feinberg AP†. A functional DNA methylation signature of acute myeloid leukemia stem cells. (Under Review in Nature Communication)

Schlaeger TM, Daheron L, Brickler TR, Entwisle S, Chan K, Cianci A, DeVine A, Ettenger A, Fitzgerald K, Godfrey M, Gupta D, McPherson J, Malwadkar P, Gupta M, Bell B, Doi A, Jung N, Li X, Lynes MS, Brookes E, Cherry AB, Demirbas D, Tsankov AM, Zon LI, Rubin LL, Feinberg AP, Meissner A, Cowan CA, Daley GQ. A comparison of non-integrating reprogramming methods. Nature Biotechnology. (2014)

Sinha S, Thomas D, Yu L, Gentles A, Jung N, Corces-Zimmerman MR, Chan SM, Reinisch A, Feinberg AP, Dill DL, Majeti R. Mutant WT1 is associated with DNA hypermethylation of PRC2 targets in AML and responds to EZH2 inhibition. Blood. (2014)

Kim K, Doi A, Wen B, Ng K, Zhao R, Cahan P, Kim J, Aryee MJ, Ji H, Ehrlich LI, Yabuuchi A, Takeuchi A, Cunniff KC, Hongguang H, McKinney-Freeman S,Naveiras O, Yoon TJ, Irizarry RA, Jung N, Seita J, Hanna J, Murakami P, Jaenisch R, Weissleder R, Orkin SH, Weissman IL, Feinberg AP, Daley GQ. Epigenetic memory in induced pluripotent stem cells. Nature 467:285-90. (2010)

Jung NY, Bae WJ, Chang JH, Kim YC, Cho Y, Cloning, expression, purification, crystallization and preliminary X-ray diffraction analysis of the central Zn-binding domain of the human MCM10 DNA replication factor, Acta Crystallogr Sect F Struct Biol Cryst Commun.(2008)

212